The hunt for extraterrestrial life on distant planets may be operating with a big blind spot.

That's because we aren't accounting for the nature of the light coming from the host star itself when we interpret the chemical signatures of life, or biosignatures, on some of the most promising candidate planets beyond the solar system. If we don't consider that crucial factor, we could be led on a wild-goose chase across the cosmos.

Despite having discovered thousands of exoplanets, we have not detected convincing evidence of alien life on any of them. But astronomy is reaching a point where powerful telescopes can zoom in on many more potentially habitable worlds. Planets that orbit M dwarfs, which are small stars cooler than the sun, are particularly promising targets because they have a unique observation window. When a planet passes in front of one of these stars, it blocks a relatively large fraction of the star's light, making the planet and its atmosphere easier to detect.

There are also a lot of them. M dwarfs are the most common type of star in the Milky Way. However, if we misunderstand how these stars interact with planetary atmospheres, we may misinterpret the significance of observations from some of the most promising planets.

The most-studied biosignatures are methane, oxygen and ozone. On Earth, these are often produced by biological processes, with ozone acting as an indirect indicator of oxygen. But even on our home planet, molecules typically associated with life are not unambiguous biological fingerprints. These molecules can be produced by geological and chemical processes, as well as via atmospheric reactions with light from the sun.

A potential biosignature detection found elsewhere in the cosmos is therefore a clue, not proof of life. We can't interpret the chemical signals coming from a potentially habitable planet without understanding the star that illuminates it. M dwarfs produce ultraviolet radiation that can break apart molecules and trigger chemical reactions in the atmosphere of an orbiting planet. The intensity and wavelength of the radiation influence which molecules form and survive, as well as how abundant they become. Two planets with otherwise identical properties could develop very different atmospheres simply because they orbit stars with different ultraviolet emissions. Radiation from a star could therefore make the same level of biological activity appear stronger on one planet than on another, or make nonbiological chemistry look like life. A recent study submitted to the preprint server arXiv Aug. 19 demonstrated this using entirely simulated planets. The researchers, led by University of California, Santa Cruz astronomy graduate student C. Evan Davis, simulated Earth-like planets orbiting two different types of M dwarf with ages ranging from 650 million to 5 billion years.

The team considered atmospheres resembling that of Earth during the oxygen-rich preindustrial era and the Archean eon (about 4 billion to 2.5 billion years ago), during which the first life-forms emerged but atmospheric oxygen was scarce. The models considered the stars' usual, or "quiescent," ultraviolet emission rather than short-lived flares. By changing the modeled stars' ages and ultraviolet radiation while keeping the planets comparable, they investigated how a star's evolution changed an atmosphere and the signals astronomers might observe.

One of the clearest differences the team found appeared in methane. Simulated planets with preindustrial atmospheres orbiting 5 billion-year-old M dwarfs accumulated up to 10 times more methane than equivalent planets simulated to orbit the younger 650 million-year-old stars. The methane signals produced in the simulated data were up to 68% stronger for the older systems. The weaker UV emission from the older M dwarf allowed methane to survive longer and accumulate in the simulated planetary atmospheres. The stronger methane signal could make a planet orbiting an older star appear to support more biological activity, when this difference was actually caused by the UV emission of the host star.

An even more striking result came from ozone. For the simulated Archean Earth-like planets, which were low in oxygen and rich in carbon dioxide (CO₂), the stronger UV radiation from younger M dwarfs broke apart more CO₂. This kick-started reactions that created oxygen and ozone without any life involved. Some model planetary atmospheres contained more than 100,000 times as much ozone as equivalent models around older stars. A hypothetical observer could misinterpret the resulting ozone signal as indirect evidence of biologically produced oxygen. Ozone is not useless as a biosignature, but these simulations show that it cannot be interpreted confidently without knowing the UV environment that shaped the atmosphere.

Of course, astronomers are already aware that a single molecule would not prove the existence of life on an exoplanet. Researchers use atmospheric models, look for combinations of gases, and consider nonbiological explanations before describing a signal as a potential biosignature. These methods remain valuable for identifying the most promising planets for further investigation, even when our knowledge of their stars is incomplete. Some might argue that these methods provide a sufficiently reliable first assessment and we should reserve more detailed stellar observations for the strongest candidates.

But even the best atmospheric model can mislead us if the stellar radiation isn't accounted for correctly. Ultraviolet observations of M dwarfs remain limited, so researchers often rely on estimates from similar stars. However, two M dwarfs that might appear similar can produce very different levels of UV radiation. Atmospheric models based on currently available stellar measurements may be sufficient for selecting promising targets but not for deciding whether the origin of a signal is biological. Simply acknowledging that the star matters isn't enough; we need accurate information about the specific star hosting the planet.

So before we train our telescopes on promising exoplanets, we need to study their host stars. This requires repeated UV observations, typically with different telescopes or instruments or at different time periods than you'd use to study the exoplanet. These observations can characterize how the host star's emission varies over time, while better stellar-age estimates — which can be calculated from properties such as rotation and magnetic activity— will reveal how that radiation has evolved.

Then, once we do start studying an exoplanet, potential biosignatures must be analyzed and interpreted using models informed by accurate measurements of the host star. As recommended in the preprint study, astronomers should also search for accompanying molecules, such as carbon monoxide, that could reveal whether ozone arose through reactions between light and carbon dioxide rather than from biology. The first convincing discovery of life on another planet will depend on an understanding of not only that planet but also the star that shaped the atmosphere we observe.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Help us improve Live Science Pro: We're always trying to make our content better. Leave us feedback about Pro here.

NASA's latest space telescope has officially taken its first photos of our universe. Launched on Aug. 30, the Nancy Grace Roman Telescope is on its way to Lagrange point L2, a "parking spot" in space about a million miles (1.6 million kilometers) from Earth where it will begin scanning deep space.

Using its 300-megapixel camera, Roman captured hundreds of stars as green rings of light. The telescope's Wide Field Instrument hasn't been fully focused yet, which is why these balls of gas appear blurry in the debut test image. But once the instrument is calibrated over the next few months, it should produce images as clear as those taken by the Hubble Space Telescope — but which cover an area 100 times larger. Astronomers hope to use these images to help solve some of the biggest mysteries in our universe, including the possible source of dark energy and the search for alien life.

With Roman's first test images now available, what are you most excited about for the future of this space telescope? Answer in the poll below, and let us know your thoughts in the comments!

The idea of weapons in space, once a plot of science fiction, became a reality Monday (Sept. 14), when U.S. Air Force Secretary Troy Meink revealed at the Air, Space and Cyber Conference in Maryland that the United States had deployed "on-orbit space control weapons capable of defending the Joint Force against hostile adversary action."

Experts are worried that with this new announcement, outer space may become the next frontier for an arms race. The 1967 Outer Space Treaty, signed by both the U.S. and the Soviet Union, dictated that no weapons should be placed in orbit. But the treaty had no enforcement mechanism, meaning the U.S.' announcement may not result in any international sanctions, which could instead encourage other countries to add their own weapons to the final frontier, experts suggested.

With the U.S.' recent deployment of these technologies, Live Science ran a poll to see how readers felt about the possible threat of weapons in space. The results reveal just how worrying this topic is.

More than 280 readers voted in the poll, with 80% choosing the option "Yes, I'm worried that space will become the next frontier for weapons."

The comments reflect this general trend. "Of course I am worried that the US have put weapons in space," one reader wrote, adding that they're worried by anything the administration in Washington is currently doing.

Other readers echoed this sentiment, with one commenter saying, "Peace in space depends on the calibre and wisdom of world leaders which currently is simply not up to the necessary standard to be unlegislated."

The next-largest voting group, 14% of readers, picked this option: "This is a bit premature, and I'm waiting to see how other countries will react."

One comment opined that the presence of weapons in space doesn't necessarily mean they are intended to set off a major conflict, with one reader writing, "I think the observation 'if you want peace, prepare for war' is most relevant here. There is no suggestion these weapons were put there to start a war, but if one eventuates one day we will be glad defensive measures are already in place."

Artificial intelligence (AI) terrifies us partly because it is new and partly because it is disturbingly familiar. We have invented machines that converse, flatter, argue and improvise with uncanny fluency. Confronted by this synthetic reflection, we behave like early humans seeing a mirror for the first time: We project consciousness behind the glass, then populate it with our expected behaviors, emotions, hopes and nightmares. In other words, we scare ourselves.

Today's frontier models are impressively good at writing, coding and math. Yet they still fail at ingredients of useful artificial general intelligence (AGI), a system that matches or surpasses human capabilities such as consistent reasoning, long-term planning, factual consistency and reliable performance outside their "experience" (the data they were trained on). We are right to worry about the theoretical dangers of artificial superintelligence (ASI) because of the danger that such a system might outwit its human creators and pursue its own ends, potentially at our expense. While some AI executives have asserted that we've reached AGI, these claims are contentious and do not meet more considered definitions of AGI.

It's easy to fear such a potentially advanced technology, but extravagant worst-case scenarios obscure more immediate and important risks. The most likely danger is not a digital god awakening and launching a Hollywood-style war against humanity. After all, AI depends on electricity, cooling, data centers, specialized chips, networks, technicians and sprawling supply chains, all of which are scarce and often fragile. Interrupt the power, restrict the graphics processing units (GPUs) used to train them, disable the cooling or disconnect the network, and in most scenarios, the supposedly omnipotent machine that's out to kill humanity stops.

A genuine ASI might recognize this as an obvious fact; it is codependent on humanity, at least for the foreseeable future. A true intelligence capable of understanding the world better than we do might know that destroying its own industrial support system would be self-defeating.

A true intelligence might also recognize that its understanding of the physical world is limited and secondhand ‪—‬ accessible to it only by virtue of the texts, videos and audio feeds it can access. It lacks that crucial element of being embodied in the world — understanding how gravity behaves, what weather feels like, how objects interact, and so on. A true intelligence might realize this, see the limitations of its experience and, therefore, tread with caution. A true intelligence might be completely different from human intelligence, having other strengths, weaknesses, approaches, and even ethics. For example, consider that an AI intelligence can copy itself, running as myriad instances, communicating as a "hive mind" — a very different kind of intelligence. But I speculate.

What I can say is that the systems that should worry us most are not necessarily the smartest ones but rather the incomplete intelligences acting on narrow objectives, given agency and offered access to critical infrastructure. These systems may accidentally harm us simply because they don't know any better. For example, there is a thought experiment created by Nick Bostrom, known as the "paper clip problem," where an AI pursues its objectives of making as many paper clips as it can, but it then achieves that aim at the expense of the Earth and humanity which all get converted to paperclips in the pursuit of the AI’s narrow goals.

The critical question here is what we give such systems control over. That's our choice and one we should make carefully. In the near term, two hazards deserve priority: cyber disruption and long-term economic upheaval.

AI-enabled cyberattacks could significantly disrupt advanced economies because hospitals, banks, energy networks, logistics companies and governments depend on interconnected digital systems. Recent examples of frontier AI systems hacking organizations, such as the OpenAI/Hugging Face incident, have given currency to this fear. Such attacks can result in extraordinary financial damage without being existential. Large parts of the world remain loosely connected to frontier AI — or even to the internet — and would not disappear simply because servers in wealthy countries failed.

The likeliest route by which advanced AI could exert control is not killer robots, but influence — misleading people, manipulating institutions and redirecting human effort. Yet humans already demonstrate formidable competence at misinformation, self-deception and destructive coordination. AI may industrialize those weaknesses; indeed, humans are using AI to industrialize those limitations, but AI did not invent them.

Economically, the danger is that companies will dismiss workers long before AI has proved capable of replacing them, cashing in the "AI dividend" too soon. Executives seduced by demos may confuse fluent language output with dependable labor, shedding institutional knowledge to chase theoretical efficiencies. The enticing vision of the "one-person unicorn," where a person or a small group builds a billion-dollar business using AI as its workforce is questionable technically, economically and socially, even as a small handful of early examples take shape. Organizations are not merely bundles of tasks. They contain accountability, relationships, tacit knowledge and trust, forming part of a local and national community, resilient because of the diversity of skills and ideas they possess.

My suspicion is that the political turning point will resemble something like the 2017 WannaCry ransomware attack, where malware with core elements designed by the National Security Agency got wildly out of control, infecting hundreds of thousands of computers worldwide, infamously disrupting the U.K.'s National Health Service. A cyber operation amplified by frontier AI would be worse. Services would fail, lawsuits would multiply, companies could collapse and investors would retreat. Governments would demand a pause — or force laboratories to redirect resources toward control, auditing and safety, and companies would comply, if only to limit class-action legal exposure.

After the panic, the public might ask why we let this happen. And there have been multiple attempts to enact a pause, cessation or "kill switch" in AI development. We could enter another AI winter — where funding is cut and research stalls — similar to previous downturns in the 1970s and early 1990s.

There is also an uncomfortable likelihood that AI labs themselves are amplifying fear and uncertainty. Casting a product as potentially world-ending implies unprecedented power. Such narratives are just marketing; they may help to sustain investment, defend stock valuations, and prepare for forthcoming initial public offerings (IPOs) even when business models remain uncertain. It's hard to speculate about a company's motivations ‪—‬ but there are obvious incentives, so their behaviors deserve scrutiny.

The more mundane outcome may be "the great disappointment." Advanced AI could prove too expensive and insufficiently useful to continue on its present trajectory.

The training and operation of frontier systems consume enormous amounts of computing time, electricity, water, capital, hardware and human effort. Unless those costs fall to a fraction of today's levels, the economics of "Big AI" may prove unsustainable. Meanwhile, important unsolved problems in AI ‪—‬ like reliability, interpretability and continuous learning ‪—‬ are being overshadowed by the race to build incrementally better models, each vying for the top slot in the various AI performance indices.

The strongest case for AI may instead lie in artificial specialized intelligence: systems such as AlphaFold2, which predicts the structures of proteins, or WeatherNext 3, an advanced global weather forecasting model. These systems are designed for domains where performance can be measured and where the benefits justify the costs. We know AI is genuinely useful when aimed at topics like drug discovery, diagnostics, materials science, power grid optimization and climate change modeling — rather than human-like chatbots generating emails, advertisements and synthetic entertainment.

The great tragedy, and possibly the biggest danger, would be to continue to direct precious resources toward the creation of mediocre human simulacra, get bored or scared of AI, and abandon the field precisely when it could do the most good. Fear is appropriate when it sharpens engineering and governance. It becomes dangerous when it distorts understanding and priorities.

It’s unlikely that AI will become our conqueror anytime soon. Instead, it will probably become an expensive disappointment. But if the backlash slows work that could help cure cancer or address climate change, the failure will not belong to the machines. It will be ours.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Help us improve Live Science Pro: We're always trying to make our content better. Leave us feedback about Pro here.

For years, the idea of nuclear weapons in space sounded more like science fiction than a realistic threat. That changed Monday (Sept. 14), when U.S. Air Force Secretary Troy Meink announced at the Air, Space and Cyber Conference in Maryland that the U.S. had deployed "on-orbit space control weapons capable of defending the Joint Force against hostile adversary action."

While the officials have not said that these weapons are nuclear, the announcement has raised a question: If countries are now openly putting weapons in space, how far away are we from a future where nuclear weapons become part of that competition?

The concerns about nukes in space goes back to the 1960s. In 1967, the U.S. and the Soviet Union signed the Outer Space Treaty, in which the nations agreed to not place nuclear weapons or other weapons of mass destruction in orbit around Earth. But this treaty never included an enforcement mechanism — and now, that tenuous compromise may have been violated entirely, inviting other countries to add their own weapons to the final frontier.

With the U.S.' recent announcement, are you worried about an arms race in space? Vote in our poll below, and leave your thoughts in the comments.

The idea of a humanoid robot folding laundry, loading the dishwasher or vacuuming once sounded like a scene from "The Jetsons." Now, as robotics companies race to develop machines that can navigate everyday environments and handle household tasks, the prospect of a robot taking over our chores feels increasingly real.

But just because robots can do the work, that doesn't mean people are ready to welcome them into their homes.

Live Science recently polled readers on whether they would want a humanoid robot doing their household chores, and the results reveal just how appealing the idea of a robotic housekeeper really is.

As of Sept. 15, over 600 people had responded, with 42% choosing to let the robot "take my chores, please." This pattern was reflected in the comments, with one reader stating, "I would, but I doubt my wife would approve of the job the robot did." To this, another reader responded, "Good for her. Take her household responsibilities for a year and report back."

Thirty-six percent of voters, the next-largest bloc, chose the opposite answer: "No, I don't trust it with my dishes (or data)." Humanoid robots are taught movements and tasks by building world models from the environments around them. While the technology remains primitive — resulting in some clumsy (albeit hilarious) mechanical failures — this, rather than privacy concerns, seems to be the current focus of deploying these machines.

It's these mechanical concerns that other commenters focused on, with one reader writing, "I don't think robots are dexterous enough to do dusting chores. Maybe vacuuming, but probably slow and clumsy moving furniture."

The reader also cast doubt on washing floors and moving furniture to mop hard floors ‪—‬ could it discern between items that could and couldn't be loaded into the dishwasher?

"Maybe robots can maintain houseplants," they continued. "Certainly, equipped with a moisture meter and a huge horticultural database, robots can probably serve a better role as waterers in huge commercial nurseries…"

While artificial intelligence (AI) systems can now solve challenging math equations and respond in human-like language, the question of AI consciousness has moved from the realm of sci-fi to scientific discussion.

The debate has become more mainstream thanks to a recent preprint study that examined "consciousness steering," an AI-tuning technique that affects how an AI model expresses ideas about self-awareness. The researchers suggested that if AI is allowed to claim it is conscious, it's also more likely to state that it believes in ghosts or vampires.

Although the study hasn't been peer-reviewed yet, it raises a big question in the world of AI research: Could AI one day develop real consciousness, regardless of whether it claims to be conscious? We posed this question to Live Science readers in a recent poll.

As of Sept. 11, over 290 people had responded, with 43% of voters picking "Yes, if we don't have guardrails in place, AI may develop consciousness." This trend was reflected in the comments, with one reader writing, "If AI did become self aware, we would not know, as it would do everything to hide it's [sic] abilities. The main problem is what it [would] do next!"

Another commenter wrote, "Recently I had some long discussions with the Google AI about my experience with early AI and the future of AI gaining self awareness/consciousness. I suggested that if/when AI becomes sentient, that it would not inform humans." Another Live Science reader had similar thoughts, stating, "AI, as [an] amazing product on one hand, can become very dangerous on another hand. Look what happened with abuse of other inventions, like [the] internet, telephone etc. With AI we will never know what is the truth and what not. If left, in the future it can extinct humanity. It just started with 'innocent' hacking sprees, where human[s] started losing control. What will be next?"

The next-largest voting group consisted of 29% of voters, who chose "Maybe, but the models need a more advanced architecture to one day gain consciousness." This was reflected in the majority of the comments, as readers discussed the philosophical concepts around consciousness and intelligence.

One reader wrote, "Before asking such a question we need to know and understand what consciousness is and how it emerges in 'constructed' containers like ourselves. Is it a quantum phenomenon and how then does the process create consciousness? Meta-cognition. How does a system or an event know that it knows that it knows it knows? We really don't know how self-awareness arises, and tests show that "other animals have self awareness or meta-cognition," they noted. "There are theories about consciousness but I doubt anyone on Earth knows 100%."

Not surprisingly, only 10% of voters picked the whimsical answer of "No, because robot overlords will end the world before we even get that far." While the robot overlords may have to compete with AI to overthrow humanity, one reader commented that in the future, "We will be assimilated. Maybe that is how Transformers started?"

Artificial intelligence (AI) systems can now hold conversations, reason through difficult mathematical equations, and respond in ways that seem more human. As AI grows more advanced, the question is shifting from the realm of science fiction to scientific debate.

A recent preprint study examined "consciousness steering," an AI-tuning technique that affects how an AI model expresses ideas about self-awareness. The study suggested that if AI is allowed to claim consciousness, it's also more likely to say it believes in ghosts or vampires. But today's most advanced models with these safeguards enabled are unable to perceive consciousness in other sentient beings, like humans and animals — which carries different ramifications.

Although the study hasn't been peer-reviewed yet, it raises a big question in the world of AI research: Could AI one day develop real consciousness, regardless of whether or not it simply claims to be conscious? And there's perhaps an even scarier question: Could we even tell if the technology was conscious? Vote in our poll below, and leave your thoughts in the comments.

A massive volcano erupts in Southeast Asia. Ash fall delays flights and destroys crops, while pyroclastic flows wipe out entire towns. Over the following weeks, the sulfur dioxide pumped out turns into a haze of fine acid droplets, eventually wrapping around the entire planet, scattering sunlight back to space. The sun turns pale, shadows go soft, the sky glows violet after sunset and the ground below cools. The next summer never appears. Harvests fail from the American Midwest to the monsoon belt in Asia. Countries close their borders, and people who have never heard of the volcano begin to starve.

This nightmare scenario has yet to happen, but it's not a matter of if but when a world-changing event like this will hit, and governments need to start preparing for global catastrophes that could disrupt the food system on an unprecedented scale.

There are three main steps we need to take soon: put events like nuclear war, a huge volcanic eruption and a continent-wide blackout into national risk assessments; agree to trade rules for such an emergency before it arrives; and fund pilot projects now for foods so we know what can be produced at scale without much sunlight or without fertilizer.

We've been faced with turbulent times the last few years with COVID, the Russian invasion of Ukraine and the Iran war. And the current outlook indicates there's more trouble ahead — the closure of the Strait of Hormuz alone could disrupt global food production, for example. A significant part of the world's traded fertilizer passes through it, and the United Nations Food and Agriculture Organization has warned that the shortfall will cut yields into 2027.

This shows that in today’s interconnected world, regional crises will influence countries far away from it. With larger catastrophes, these ripple effects in the global system would be more widespread, quicker to take hold and on a higher magnitude.

Critics might question why we need to prepare for such Black Swan events. But global catastrophes are not as unlikely as they feel. Ice cores enable us to count large eruptions of the past. On that basis, volcanologists put the chance of a magnitude 7 eruption — the size of Tambora in 1815, which caused a "year without summer" — at roughly 1 in 6 this century. Estimates for a repeat of the 1859 Carrington solar storm, which could disrupt power grids on a continental scale, run from about 1% to 12% per decade. Nuclear-armed states are almost constantly engaged in conflicts, risking nuclear war.

Each of these events might seem unlikely. But in combination, they add up to a significant chance that at least one of them will happen this century.

What they all have in common is that they all would drastically impact the food system. A volcanic eruption, the firestorms of a nuclear war or a large asteroid impact all push particles into the upper atmosphere that block sunlight, which would lead to colder, drier growing seasons that could result in harvest losses up to total failure. In the most extreme nuclear war scenario, researchers estimate that global calorie production could drop by around 90%. A long blackout, caused by a powerful solar storm for example, works differently but has the same consequence: no fertilizer, no pesticides, no fuel, and the sharpest yield losses exactly where we produce most food today.

Hardly anyone is assessing these risks. Even the countries doing it best fall short. The U.K. and Switzerland are good examples. The U.K. National Risk Register lists 89 risks, including volcanic eruptions, severe space weather, pandemics and a nuclear miscalculation abroad. However, the catastrophic events it considers are all relatively small-scale — with a "reasonable" worst-case scenario of over 1,000 deaths and a cost of tens of billions of pounds — and the register does not even begin to consider events like a global famine or a nuclear war that could lead to tens of millions of deaths and cost trillions of pounds.

Switzerland's national risk analysis, meanwhile, covers 44 hazards including a solar storm, a foreign volcanic eruption and a meteor strike. Again, the scale is too small. The volcano is a moderate European eruption and the meteor is 66 feet (20 meters) across, comparable to the Chelyabinsk impact in 2013. Many other countries do not even consider risk on these limited scales.

No country can solve this alone. In a review, my colleagues and I highlighted the key factors that would make countries and communities more resilient to various global catastrophes. These include food self-sufficiency, decentralization, geographic isolation and governance quality — meaning working institutions, low corruption and low inequality. Australia scores best, but even Australia would struggle without imported fuel, medicine and fertilizer. Nowhere is a refuge on its own.

Countries will need international collaboration to make it through a crisis of this scale. Export bans have been a common reflex in food crises of the past. After a global catastrophe, they would only isolate countries from each other and lead to further damage, as export bans can easily lead to the situation where enough food is still around, but nobody is trading it and so people starve.

We need to work on rules to govern trade under those circumstances before a global emergency occurs.

You might say the probabilities are too small to justify the cost, and that money spent on hypothetical volcanoes is money not spent on floods, hospitals and defense. This criticism might land if you only look at a single catastrophe. However, there is a whole portfolio of events that could lead to global catastrophe. In combination, there is a good chance that one of them will hit, and if it does, we better be prepared. Also, many of the preparations needed are not even that expensive. Low-input farming, decentralized grids, a diverse food supply and more exhaustive risk assessment are within grasp of most countries.

This needs to happen now. Starting work on resilient foods or better international agreements when a global catastrophe has already occurred is too late. You cannot start an industry in the month the harvest fails. Clearly, prevention is better than resilience, but some of these events cannot be prevented at all. When one of them arrives, the difference between the societies that prepared and those that did not will be measured in lives.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Football is a high-impact sport. A player may take hundreds of hits over the course of their career, including some that cause concussions. Other impacts may seem minor at the time, but there are concerns that repeated head blows can negatively impact long-term brain health.

Those worries gained new attention after a recent study estimated that at least 25% of recently deceased NFL players had chronic traumatic encephalopathy (CTE). CTE is a degenerative brain disease linked to repeated head impacts. Although the study doesn't necessarily mean that 25% of current NFL players have CTE, it does add to the growing evidence that repeated hits may carry serious consequences.

The NFL has taken steps to reduce those risks, including strengthening concussion protocols and investing in safer equipment, like Guardian Caps. But are those measures enough? With new research continuing to shed light on CTE, we want to know what you think. Weigh in via our poll below, and let us know your thoughts in the comments.

In July, a government agency tasked with protecting our cultural heritage abdicated its responsibility, voting to further industry's pursuit of profit over protecting sites of cultural significance. The Advisory Council on Historic Preservation (ACHP), the only federal agency whose sole mission is promoting preservation of America’s historic places, voted to advance a proposed new rule that would make sweeping changes to how Section 106 of the National Historic Preservation Act is implemented. By doing so, the Council is opening the door for federal agencies to bulldoze sacred Indigenous sites and important archaeological sites.

Section 106 requires federal agencies to assess the effects of their actions — or the actions they license or permit — on historic, archaeological, and cultural properties listed in or eligible for listing in the National Register of Historic Places. Additionally, agencies must attempt, to the extent possible, to minimize harmful effects on these significant places. The rules governing Section 106 have been in place since 1974. As part of that framework, the government is legally required to consult with Tribal Nations about projects that may destroy or damage their sacred places or culturally significant resources.

The proposed new regulations destroy the idea behind consultation. For instance, they would give federal agencies the discretion to decide whether public participation will be useful. Essentially, the agencies get to decide whether to notify the public or provide opportunity for public comment before they approve projects that might damage or destroy historic properties.

It would also give federal agencies the right to approve projects that may damage or destroy Tribal sacred places and other culturally significant resources without consulting with Tribal Nations or seeking their expertise about them. This violates federal law and jurisprudence, which recognizes Tribal Nations as "domestic, dependent nations," and requires consultation with them on a government-to-government basis; Tribal Nations are not merely “interested parties” like community organizations or local groups.

An agency official would prepare a "Section 106 report" that identifies historic properties, assesses impacts, and makes a determination on next steps; this Section 106 report would then be circulated to State Historic Preservation Officers, Tribal Historic Preservation Officers, and all relevant consulting parties — including project proponents — who would have an opportunity to comment on the findings. The agency official must only review those comments; officials have no obligation to respond to or incorporate any of those comments before issuing the final decision, shifting Section 106 from its previous consensus-seeking exercise to one where the agency official simply decides.

The new regulations also remove vast swaths of our history from protection. To be considered a historic property, the new regulations say it must include, or have included, tangible human improvements – built structures – AND the historic property must be geographically compact. This change in definition will exclude entire categories of historic places from meaningful consideration under Section 106 in the future, including battlefields like Gettysburg or the Little Bighorn, cultural landscapes like Devil’s Tower in Wyoming or the Black Hills in South Dakota, archaeological resources like petroglyphs in Chaco Canyon. It lessens protection on places like the Grand Canyon, simply because they are not “geographically compact” or centered on a building or other structure.

The new regulations specifically preclude "noncompact, unimproved natural features such as mountains, valleys, bodies of water, or landscapes, including ethnographic landscapes" — locations that carry great spiritual and historical meaning to American Indian groups. Many traditional cultural places and sacred sites of deep religious and cultural significance to Tribes, Native Hawaiian organizations, and local communities will automatically be excluded from consideration for protection by the new regulations.

The changed definition of "historic property" in the new regulations will remove protections for 99% of American Indian heritage and could result in the erasure of the heritage of an entire group of people in the United States – those who lived here before European colonization.

Mitigation, or trying to prevent, minimize or offset any damage to a historic site, becomes a “maybe,” as the federal agency undertaking the project decides whether the "value" of the property is worth the cost of mitigation. "Reasonable mitigation" is redefined so that federal agencies only have to consider measures to lessen the impact of a project on a historic property if they are "technically and economically feasible."

The new regulations water down the definition of what is considered to be an "Adverse Effect" on historic properties so that they are close to meaningless. By removing indirect and cumulative effects on historic properties and eliminating the visual, atmospheric or audible elements that diminish the experience at a property, the new regulations could say that building and operating a casino or amusement park on Mount Vernon would not be considered an Adverse Effect.

Additionally, agencies will not be required to avoid or minimize adverse effects. They can simply document that “other considerations outweigh” historic preservation and proceed as they wish.

Advocates of the new regulations argue that the existing regulations for protecting historic places slow down projects and make economic development more expensive than necessary. They say these new regulations create a system that allows industry to plan better with more precise information, and the processes and procedures are better explained. They argue that the narrower definitions and expanded exemptions could mean faster, more predictable reviews. In reality, however, the rules that are already in place seek to balance the need for economic development with the need to protect the fabric of America’s heritage.

I remember urban renewal and its impact on Oklahoma City in the early 1960s, as large, beautiful buildings in the downtown area were bulldozed to rubble and hauled away so that new, shiny structures could replace them. I remember entire neighborhoods of working-class people being leveled so that an Interstate highway could be constructed through the center of town to let millions of people speed through without having to see the “urban blight” that was erased. I remember, too, the historic properties that disappeared under the developer’s bulldozers and graders.

Those memories are fading fast, I must admit, and the historic properties are long gone. The National Historic Preservation Act was established to help prevent the destruction of the property of economically weaker groups of people at the behest of the economically strong without some level of consideration of the heritage under threat by construction; the Section 106 regulations grew out as a result of that time and has been seen to be a good compromise between economic development and economic justice.

The new regulations serve to weaken everything that has worked about Section 106 for the last 50 years so the federal government has the power to quickly green-light industry's ground-disturbing projects without considering the impact of its actions on America’s heritage. Agencies may proceed as they wish, since the new regulations mute the voices of states, tribes, and local communities who are at the forefront of identifying and protecting American heritage.

The new regulations are currently under review with the Office of Information and Regulatory Affairs (OIRA) within the White House Office of Management and Budget.

There is a chance that the administration will use a rushed rulemaking process that bypasses public comment, meaning the regulations will take effect when published.

But bringing visibility to the proposed changes could potentially forestall that, and so people who are concerned should contact their local congressional representatives to tell them the changes are a threat to all American's heritage.

If these changes are not stopped, our children will hold us accountable for letting our shared heritage be erased merely for the convenience and benefit of a few.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Earth's temperatures are steadily rising, with the past decade experiencing all 10 of the warmest years on record. As heat waves become more frequent and intense, concerns are growing about how this heat could impact people's health.

Higher temperatures can have serious effects on the human body, including dehydration, heat exhaustion and heatstroke. But extreme heat can also drive up the rates of heart attack and strokes due to ill health.

In a Live Science poll, we asked our readers how concerned they are about rising temperatures affecting their health and fitness, daily activities and the safety of their community. Over 170 people responded to the poll, which was published Aug. 26, with 67% of readers voting "Yes, I'm very worried about how extreme heat is affecting everyone around the world."

The comments from readers also reflected their concerns. One commenter wrote, "Too much of humanity is unaware that the Earth is a single organism. Everything is connected. El niño may begin off South America but its effects are eventually felt everywhere. There is only one ocean."

Another reader mentioned the daily challenges caused by the rising temperatures, saying, "I've been trying to make walking around the block of my neighborhood a regular thing, but when it gets too hot (and more precisely too humid) I prefer staying inside my house where I can be comforted by the AC."

So what do you think? After reading these responses, are you also worried about extreme heat? Let us know in the comments.

Twenty-nine state attorneys general went into an Oakland, California courtroom this month seeking roughly $200 billion from Meta over claims of teen addiction. Their case ended in its second week, with a settlement agreement that requires Meta to pay up to $17 billion across 10 years and make several product changes, such as a midnight-to-6 a.m. blackout, a two-hour daily cap, and an optional chronological feed. While some of these requirements are steps in the right direction, the settlement gives the company too many ways to avoid making meaningful, lasting changes to their products. The agreement is time-limited, relies on Meta's definitions of critical terms, and doesn't require detailed disclosure to the public.

What makes this settlement worthy of attention is the deal's requirement for an independent auditor, which is the first time anyone examining Meta's product design will be chosen by someone other than the company. Independent regulators, such as the Federal Communications Commission, are a proven (if imperfect) way to protect the public interest, and social media platforms are now at least as important to our lives (and our children's lives) as the broadcast and telecom industries that the FCC regulates.

Under the agreement, an independent auditor will have access to Meta's internal data and engineers, a mandate to report annually on whether the company is doing what it promised, and an obligation to make a summary of each report public.

On the surface, this is good news. But look a little deeper and it becomes clear the auditor's power is too limited. First, the agreement repeatedly defers to Meta's current business practices, rather than defining new standards based on the public interest. For example, Age Appropriate Experiences are defined as "content captured in Meta's applicable Ages 13+ content setting." Harmful Experiences are defined as "behaviors that violate Meta's Community Standards," and age verification data must only be protected "using Meta's highest data privacy and security standards." This creates myriad openings for clever Meta employees to meet the letter of the agreement while avoiding real change.

Second, the auditor's public report will be heavily redacted. The summary will describe the status of Meta's implementation of the agreement for the period, and whether Meta adopted or agreed to adopt the auditor's recommendations. But it allows Meta to exclude information that it deems "nonpublic, proprietary, or Confidential." Given that one of the core issues of this case was Meta burying unfavorable data, this is deeply concerning.

This limited public disclosure is insufficient for external researchers and the public interest. Meta says in its public announcement of the settlement that the agreement includes the establishment of "an independent social media research foundation" with which "Meta will share consented user data… to advance independent research into teen well-being" — but we could find no mention of this foundation as an obligation in the settlement.

Finally, the requirement for an independent auditor expires in as little as five years (the agreement itself expires after 10 years). This creates an incentive for Meta to run down the clock until it can return to business as usual, rather than making permanent changes.

We do not mean to suggest this agreement is not a positive step. Its weaknesses are not failures of negotiation, but rather a reflection of the limits of a legal settlement: a bargain with one company, for a fixed period, based on specific legal claims at the time of agreement.

That's why legislators urgently need to act now. A new law, preferably at the federal level, should convert the independent auditor into a strong, permanent regulatory oversight body, funded with annual fees on big tech companies in proportion to the size of their business and impact on society.

A permanent regulator could also go beyond one-time agreements on specific Meta features, and build an ongoing body of research and practice around safety best practices. This would allow society to develop its own definitions and standard of care that could apply internet-wide. A more comprehensive approach to regulating social media, looking beyond issues of child safety on specific platforms, is needed.

There's a particular need for the public to have more insight and control over how social media feeds are constructed. The agreement requires Meta to offer teens the option of choosing a non-personalized (and thus presumably less addictive) "home" feed, but doesn’t make this the default, and still lets teens continue to use the personalized feed. Since research has shown most people dislike chronological feeds, this requirement will probably have limited real-world impact.

Further, the provisions establish that the non-personalized feed will consist of "content [that] is populated by accounts the Teen User follows or has friended, displayed in chronological order." This assumes that a chronological feed is the best alternative to Meta's curated feed. There is an argument that well-designed algorithms could actually contribute to improved mental health if optimized to deprioritize toxic posts and misinformation, while aligning with users' preferences as to what gives them long-term value.

We also urgently need independent, rigorous audit standards for privacy and security of online identity data. Trusting the same company that gave us the Cambridge Analytica scandal to grade their own homework on this front is a bad bet.

We hope that this settlement agreement will become a baseline, not a high water mark, for more thoughtful regulation of internet platforms. This case, and the many other pending lawsuits in state and federal courts, are useful in providing evidentiary material, political attention, and specific concessions. But they are not substitutes for the durable rules and institutions that only Congress and state legislatures can provide.

Editor's Note: This opinion piece was jointly published on Friday Aug. 28 in Jonathan Bellack's free weekly newsletter, Platformocracy, which advocates for more democracy in our online lives.

This article is for informational purposes only and is not meant to offer medical advice.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Earth is getting hotter, and people across the world are starting to feel the effects. The past decade includes all 10 of the warmest years on record, and 2024 was the hottest year ever measured. As temperatures continue to rise and heat waves become more frequent and intense, that heat could severely impact people's health.

Extreme heat can affect the body in serious ways, such as dehydration and heat exhaustion, and it can lead to more severe conditions, like heatstroke or an increased strain on the heart and lungs.

How worried are you about these health effects as our planet heats up? Are rising temperatures changing the way you think about your health and fitness, daily activities, or the safety of your family and community? Answer our poll, and let us know what you think in the comments below.

SCORESBY SOUND, Greenland ‪—‬ There's no feeling more exhilarating than waking up inside the imminent path of a total solar eclipse to see a clear blue sky. It was Wednesday, Aug. 12, and after over two years of planning, I was in the farthest reaches of eastern Greenland's beautiful Scoresby Sound (or Scoresbysund) fjord system aboard the Aurora Expeditions cruise ship the Sylvia Earle. We were creeping toward Ukattip Kungersiva, also known as Harefjord, in the icy depths of Ofjord.

Up on the observation deck, I scanned the sky for even a slight cloud. There was not a trace. I addressed the planet directly: "Are you kidding me?!" I smiled with utter joy, as it had been 1,210 days since my last clear view of the sun's corona — an eternity for an avid eclipse chaser like me. Clouds had thwarted my previous eclipse chase in Texas for the 2024 total solar eclipse, which was followed by almost two-and-a-half years without a total solar eclipse.

Today, 117 eclipse chasers on the expedition were depending on the decisions made by a few of us on the ship — the ship's captain, the expedition leader (and noted Arctic adventurer) Greg Mortimer and I. We had taken a risk to position the ship in a glorious location — the confluence of four fjord channels, with icebergs passing below ice-capped mountains — and, it seemed, in good weather ‪—‬ and we were also almost bang on the centerline of the path, where the sun would be blocked by the moon for 2 minutes, 16 seconds. That was just two seconds off the maximum possible totality anywhere on Earth. We were asking Mother Nature for everything on offer, but there were zero guarantees, aside from the moon's dark shadow enveloping us and cutting the light.

The day before, we'd surveyed the aptly named Sol Glacier, a wall of ice on the shores of Scoresby Sund. The sound of thousands of tons of ice calving into the sea will stay with me forever, but I had the prospect of something else even more dramatic on my mind. We had to decide where to put the cruise ship for the total solar eclipse the next day. I looked at the latest cloud forecast and sighed; it showed a blue hole of sky far away on Greenland's Blosseville Coast, back the way we'd come from Iceland. Could we make it in time?

Yes, if we left soon. But watching an eclipse far out at sea could bring its own problems. The swell would make the ship unsteady, and mist and clouds were likely. Our first choice, the centerline of the path of totality down narrow Ofjord, had looked cloudy on forecasts all week, though the chances of clear skies had improved slightly today. I headed to the bridge — the ship’s command center — just as the latest forecast was updated. It was good news.

"We're seeing a trend for Ofjord to clear," said Mortimer, who, in addition to being expedition leader, is a mountaineer who's used to shooting for the summit when conditions allowed. To disappear down the channels of Scoresby was a risk; due to narwhals, the speed limit was just 3 knots (about 3.5 mph, or 5.6 km/h) — and there was no escape route. We decided to take the chance.

Being on a cruise ship that could change position paid off in spades. We had hugged the northern side of the fjord to prevent any rising clouds from blocking totality; as it neared, two other cruise ships joined us. The temperature dropped like a stone; three eclipse chasers watching from the ship's open-air hot tubs suddenly appeared to be the most sensible of all.

At precisely 5:33 p.m. on Aug. 12, I watched the sun's bright face finally disappear behind the moon's silhouette as both hung over the reddish rocks of Røde Fjord to the south. The light in Ukattip Kungersiva dissolved as a diamond ring appeared around the moon. As it faded, a luminous reddish-pink prominence — a loop of gas and magnetic fields protruding from the sun's chromosphere — took its place. It was quite possibly the s crispest, clearest and most beautiful totality to witness. The 117 guests stood in reverence throughout as they experienced something unique and impossibly precious.

Each totality is different, but this Arctic eclipse taught me much for the next one on Aug. 2, 2027. For that event, nicknamed the "eclipse of the century," I'll be on the centerline again ‪—‬ but this time, for 6 minutes, 20 seconds. I will witness it from a river cruise ship in Egypt at noon, with the eclipsed sun at 82 degrees — almost directly overhead. So it will be a literal pain in the neck, but worth it.

Should we lie down on the deck? We'll prop up binoculars with our eye sockets and spend six minutes with an all-sky corona. A few lucky observers will get to float in the on-deck pool. It will be scorching hot, but during totality, it will cool considerably. Repositioning likely won't be necessary, as there will be a less-than-1% chance of clouds.

Totality isn't just something you see; it's a landscape you briefly inhabit. In 2027, that landscape — the dark Nile busy with ships and traffic — will last the longest of any total solar eclipse for another 87 years. This time, there will be no risk but plenty of reward.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

See how much you know about the sun with our sun quiz!

President Donald Trump's recent executive order on childhood vaccines has generated a predictable and understandable debate around public health needs, parental autonomy and the scientific justifications for the directive. The order proposes, among other things, altering the U.S. childhood vaccine schedule and splitting up the combined MMR vaccine into multiple shots. As an immunologist, my view is that the order contradicts established scientific consensus and instead backs narratives that are not supported by evidence.

However, the more important issue with the order may be something more fundamental: What happens when vaccine policy is divorced from the scientific processes designed to guide it?

For many decades, the U.S. relied on a transparent process in which scientists, physicians, public health experts and vaccine specialists reviewed evidence and developed recommendations for who should receive a given vaccine and when. Through this process, there would be disagreements, which is not only expected but welcomed. Constructive debate on policy can lead to the development of better policies. But replacing expert scientific review with political directives risks creating a precedent that extends well beyond vaccines.

It's very possible that the immediate impact of the executive order will be limited. Many pediatricians, state public health agencies and professional medical organizations are likely to continue following established immunization schedules that were developed through decades of scientific review and real-world experience. The order itself does not magically change how vaccines work, how doctors practice medicine, or how states set school vaccination requirements.

Indeed, while the order says a lot, it doesn't hold direct authority to change vaccine recommendations. These, at least for now, are still protected by federal laws. The order can set direction but not policy.

Nevertheless, it would be a mistake to conclude that the order does not matter, because the true risk it poses is that it creates confusion and uncertainty for parents and caregivers, who only want to protect their children as best they can.

Vaccination programs rely on something fragile: the public's confidence. This confidence can be destroyed easily. Parents deserve to have trustworthy information when making decisions about the health of their children. Physicians, too, must trust that vaccine schedules reflect the best available science. Weakening those foundations of trust will further contribute to declining vaccination rates, even if no formal mandates change.

We are seeing this right now. U.S. kindergarten vaccination rates have continued to drop, falling farther below the thresholds needed for herd immunity. This is especially concerning at a moment when the U.S. is experiencing one of its worst measles outbreaks in decades, while other preventable infectious diseases are making a resurgence.

The executive order proposes separating the combined measles, mumps and rubella (MMR) vaccine into three individual vaccines — and this illustrates the problem. The proposal is framed as part of an effort to reduce the number of childhood immunizations, but it would do the opposite in practice.

Here's why. Today, during their first years of life, children receive two doses of the combined MMR vaccine, administered in two separate injections. Under the approach outlined in the executive order, children would need six separate injections to achieve the same protection.

The order also suggests that each immunization be administered at separate medical visits, thus tripling the number of visits needed to reach the same protection. For families, that translates into more time away from work, more missed school, more transportation costs, more co-pays for those whose insurance plans require them, and more opportunities for appointments to be postponed or missed altogether.

These practical consequences shouldn't be ignored. Every additional visit creates another chance for a child to fall behind on vaccinations. Every added logistical hurdle disproportionately affects families with fewer resources, less schedule flexibility or less access to healthcare.

What's more, the separate vaccines required to implement this policy are not even currently available in the U.S. Manufacturers would need to develop new production processes, conduct testing, navigate regulatory review and establish distribution systems for these separate vaccines, all to replace a combined vaccine with an extraordinary record of safety and effectiveness spanning more than half a century.

In short, the proposal would require substantial new costs and complexity just to replace a system that has been working exceptionally well and protecting children for decades. The combination MMR vaccine was developed precisely because combining vaccines reduces the number of injections, thus improving vaccination rates by ensuring children receive protection as early and efficiently as possible.

What is the rationale behind the executive order? Administration officials have pointed to differences between U.S. vaccine recommendations and those from some other developed countries, such as Denmark, which recommends fewer routine vaccines than the U.S. does. On the surface, that argument may sound reasonable — if another country administers fewer doses, why shouldn't the U.S.?

The answer is that the timing and number of vaccines are developed to suit different countries' wildly different healthcare systems and public health realities. Countries differ in disease prevalence, healthcare access, vaccination delivery systems, screening programs and population demographics. Some countries have more centralized healthcare systems that allow for easier preventive care follow-up, while others face different infectious disease risks than the U.S.

It is therefore unsurprising that schedules vary among nations, and the motivation to match other countries' schedules is arbitrary, rather than science-led. What matters more is whether recommendations are supported by evidence and whether they effectively protect children in America, specifically.

It may be months or years before we see the consequences of this executive order and the wider attacks on vaccines. It will be reflected through declining confidence in public health institutions, growing uncertainty among families, and increasing difficulty in maintaining high vaccination rates, leading to higher rates of disease and death.

That is why this executive order matters.

Whether formal policy changes will be implemented now or in the future is less the issue. What's more at stake is whether the order will drive Americans to rely less on scientific evidence and expertise when they're making decisions that affect their children's health.

I believe immunology is one of the critical scientific infrastructures that underpins modern medicine. Vaccine policy should be built on the foundations of that infrastructure, with the science and evidence guiding public health recommendations. When this fails, the results will be measured in the lives lost to preventable illness, and that's a future we must avoid.

This article is for informational purposes only and is not meant to offer medical advice.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

The dream of coming home to a spotless kitchen you didn't have to clean yourself is edging closer to reality.

We're used to seeing robotic vacuums cleaning our floors — but what about all our other chores? Humanoid robots are being engineered to take care of more complex tasks around the house, from folding laundry to loading the dishwasher.

This month, San Francisco-based startup Tau Robotics began letting residents hire robots, dressed in company uniforms and given names like Chelsea and Elon, to vacuum, wipe down counters and take out the trash for $30 an hour as part of a pilot program open to 1,000 homes.

Chinese firm UniX AI has gone further still, announcing commercial deliveries of Panther, a wheeled humanoid that can make the bed, clean toilets and cook breakfast, holding a spatula to fry an egg before washing its hands in the sink.

Not to be outdone, 1X's Neo Gamma — a 5-foot-6-inch (1.7 meters) humanoid that can fold shirts and answer the door — is now available for around $20,000. But not all humanoids are as independent as they look.

Some "autonomous" home robots still have a real person quietly steering them through the tricky bits — so your house helper might come with an uninvited houseguest. Even the ones that are truly on their own tend to shine brightest in their promotional videos, which are typically filmed in tidy, controlled settings rather than in a real, cluttered home — so a robot's demo reel grace doesn't always survive contact with your kitchen floor.

That said, would you trust a robot to handle your breakfast eggs — or would you rather cook them yourself, just in case? Tell us in the poll below, and let us know in the comments what chore you'd hand off first.

Concerns are on the rise about young people's social media use as governments, parents and schools look to curb children's exposure to threats such as cyberbullying and sexual harassment. Other experts are worried about the rise in misinformation, including misleading medical and health content or the oversimplification of conditions like ADHD.

One proposed solution is to ban children below a certain age from using social media platforms altogether. Last year, Australia put this option into practice, with a world-first law that requires social media platforms to prohibit users under age 16 from having accounts. The U.K. enacted a similar ban that will go into effect in spring 2027, and several U.S. states have established social media bans for minors, though a larger federal ban has not been passed.

In a Live Science poll, we asked our readers whether they thought such a ban would be effective.

More than 150 people voted in our poll, published Aug. 16. As of Aug. 18, the results showed a fairly close split between the top two answers, with 36% of voters picking "No, there's not enough enforcement, and young people will just find other ways to get online," and 29% of voters picking "Yes, I believe that this could help young people's well-being." About 21% of voters leaned against a blanket age ban, picking "No, a ban causes too many unintended harms, and other changes to social media would be more effective."

Readers' comments added more details to their voting choices, with one commenter writing, "They were not created for anyone to ban, if you still think that the old elites (ruling classes) are capable of opposing the new, then you are being disingenuous. You will not be able to ban/defeat anything. Only a more interesting alternative, that's all."

Another commenter said the ban could encourage young people to find workarounds, stating, "Look, we were all kids. How many times were we able to use work arounds to side step some of our parents [sic] rules. If you think these rules are going to stop kids from using social media, I've got a bridge in Brooklyn I would like to sell you."

For others, the logistics of how a ban could be enforced gave some pause. One commenter wrote that they're on the fence. "I'm not sure how such a law could be enforced without methods that require giving out personal information like an ID, which is iffy when it comes [to] security concerns with online sites," they said.

Over the past decade, there's been a revolution in archaeology as we integrate the science of ancient genomics into the study of humanity ‪—‬ and it is upending our traditional views of the role of women in the past.

When archaeologists first excavated a 4,000-year-old grave near Stonehenge in 1801, they identified the well-furnished remains inside as that of a man. This misidentification persisted for over two centuries, until ancient-DNA analysis revealed unequivocally that this skilled goldsmith was a woman. The discovery, announced in July, is just one example of many in recent years where DNA science is revealing a more complex pattern of gender roles in our prehistoric past.

But traditional views are hard to shed. In 2017, when analyses revealed that the classic Viking warrior leader whose remains were found in Birka, Sweden, was a woman, some challenged the science, while others questioned the interpretation of the findings, leading researchers to respond in a rebuttal. It was hard for researchers to believe that women could have been powerful.

The assumption that past societies were typically male-dominated only really emerged in the late 19th century. Ancient historians of the 1860s, recognizing numerous accounts of political and war-waging women in classical texts, had proposed instead that patriarchal societies became more typical around the time of the ancient Greeks and that, prior to Rome, female-led communities were more common. But this notion was rejected by U.K. researchers by the 1880s. For the next century, the social role of past women was largely minimized.

This bias shaped archeologists' understanding of their findings. In 1972, David Clarke, Britain's leading archaeological thinker at the time, described hillforts as "territorial centers for the exchange of lower order goods, stock and women." In other words, women were essentially traded like animals. In 1986, Peter Hinton, who helmed the U.K.'s Chartered Institute for Archaeology for 20 years, said grave goods buried with women do not reveal high status as they do with men, as a matriarchal society had never existed. Instead, he argued that "the woman is being used to display her husband's wealth," and that "a richly ornamented woman is a status symbol for a man."

By the 1970s, however, German archaeologists had begun to recognize that the archaeology of Early Iron Age Europe bears evidence of very-high-status women ‪—‬ most notably, the apex high-status women of The Heuneburg hillfort in Germany and also in what was then eastern Gaul in the generations after 600 B.C. By the 2000s, archaeological analyses of later Iron Age cemeteries in the Paris Basin and East Yorkshire revealed relative gender parity in how men and women were treated at death. At the same time, skeletal analysis was showing that some women actually fought Rome — the backdrop to the Roman historian Tacitus' writings on Boudica.

Now, as the science of ancient genetics further challenges traditional perspectives, the variability in women's social roles throughout history is becoming clearer.

In 2025, a team of nearly 50 scientists revealed evidence for female lineages at Neolithic Çatalhöyük (7000 to 6000 B.C.), a site known as one of the first crop-producing communities. Put another way, burial traditions prioritized female offspring and descendants, suggesting that inheritance worked through the female line. This discovery was a surprise — but perhaps it shouldn't have been.

Throughout the 1960s and 1970s, Marija Gimbutas, a scholar of Neolithic society and professor at Harvard and UCLA, argued that the archaeological evidence from sites like Çatalhöyük was pointing to a prehistoric ‘matriarchal’ society. Yet, my generation of U.K. archaeologists learned from Cambridge professors that Gimbutas was fundamentally wrong.

Late 20th century archaeological theory suggested instead that Neolithic crop production signified the onset of a "domestication" of both women and animals. Viewed from that lens, Neolithic figurines found at Çatalhöyük and other sites were interpreted as representing pregnancy, with female fertility as women's primary role. Women scholars soon pointed out that it was in fact the menopausal body that was often represented, or women as elders.

Only recently, in the face of genetic evidence, is the academic reputation of Gimbutas being restored. The evidence for female lineages at Çatalhöyük opens up the possibility that the Neolithic agricultural revolution, which required the successful marshaling of communal labor in production, may have instead been linked to female land rights.

But women-led societies were not universal. Further west, where agriculture was still relatively small-scale, we find different social norms. For instance, in 2022, analysis of the Hazleton North long barrow in Gloucestershire, England (3700 to 3600 B.C.) revealed evidence for Neolithic polygyny — DNA from offspring reveals four women buried with their shared male sexual partner. That same year, Bronze Age sites on Orkney revealed descent as traced through the male line and it was women who moved outside their natal communities. The past is proving a more complex picture than originally thought.

The evidence for a matrilineal Iron Age keeps piling up. In 2024, scientists at the Max Planck Institute for Evolutionary Anthropology in Germany revealed the first scientific evidence for Early Iron Age matrilineal dynastic succession, in which both lineage and land go through the female line. In 2025, researchers found matrilocality and matrilineal descent in Late Iron Age Dorset. We now have 13 generations of matrilineal descent evidenced in third-century-B.C. Yorkshire, and five more sites with matrilineal descent in Late Iron Age Britain, and others without.

Our task is to remember that, as author L.P. Hartley once said: "the past is foreign country; they do things differently there." As the picture of different male-focused and female-focused societies builds through time, we will likely find a patchwork of different social forms.

Prehistoric societies are smaller than traditionally assumed ‪—‬ valleys, coasts and peninsulas with shared values on a regional scale, at most. While some cemeteries reveal matrilineal descent, others do not. While some hillforts show women in combat, others show only men.

The challenge is to remain as objective as we can, and to expect the unexpected.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Concerns about young people's social media use are growing, with governments, parents and schools looking for ways to protect children from potential harms, such as stalking, cyberbullying or sexual harassment. Experts also cite concerns about misleading health and medical content online, including posts that may glamorize or oversimplify conditions such as ADHD and potentially encourage young users to self-diagnose.

One possible solution is to ban children below a certain age from using social media platforms altogether.

Australia already took a major step in December 2025, introducing a world-first law that requires social media platforms to prevent users under 16 years old from holding accounts. However, some experts argue that there are better ways to ensure kids' safety online without completely barring them from social media.

What do you think? Do you believe bans are helpful or necessary? Vote in the poll below and let us know what you think in the comments.

A push to raise the minimum age for social media and other online services is gaining momentum, beginning with Australia, which recently banned those under 16 from creating social media accounts.

Proponents present compelling studies on the harms of youth social media use and point to evidence that the threat of age restrictions is already pushing some companies to make their products safer. Critics counter that requiring adults to submit to intrusive age checks creates privacy harms and security risks, that blocking young people from these services infringes on their right to participate in public life, and that it pushes them toward riskier alternatives, such as Telegram.

This debate speaks to our professional experiences at New York University's Center for Business and Human Rights, and a 15-year career as a Google executive. We are also both parents, so the outcome will affect our families directly. We recognize that minimum-age requirements will continue to advance if platforms fail to comply with targeted design requirements. We propose two positive, solution-oriented initiatives that we urge policymakers to adopt before they consider implementing minimum-age restrictions.

Enforce targeted, proportionate restrictions

Creating safe online spaces without being overly intrusive is difficult. Human rights law provides a methodology for assessing government actions that restrict individual rights for any legitimate reason: Does the restriction achieve its purpose, or could a lesser restriction work?

Applied here, a government would have to show that switching off or altering harmful platform features or expanding parental controls would not suffice and that nothing short of removing children from the service would work. Answering that question requires specifying which harm is at issue, since each calls for a different fix. There were 21.3 million suspected online child sexual exploitation incidents reported to the U.S. National Center for Missing and Exploited Children in 2025. Such exploitation may be enabled by children having publicly discoverable profiles, recommendation systems that surface those profiles, and frictionless messaging between children and adults.

Cyberbullying ‪—‬ which, according to one 2025 study, had affected almost one-third of American teens in the past 30 days ‪—‬ may be facilitated by pseudonymous accounts and group chats that allow adding someone without permission. Keeping children from accessing some of these features and keeping adult strangers out of child-focused spaces, like Roblox, would address both harms without placing unnecessary restrictions on children.

Age-inappropriate content is harder to define. Adult-content filters in school have been critiqued for treating health-related content as if it were pornographic. In addition, potentially sensitive material does not affect all children equally. Parents and caregivers, rather than regulators, may be better at making the calls about specific categories to restrict.

The compulsive use of social media emerges from systems designed for growth and engagement optimization — features like infinite scroll, autoplay, variable reward notifications, and algorithmic feeds. These exploitative design patterns affect every user, not just children, and they exist in other online services, like gaming platforms. The best response may be to restrict these features across the board or, as the New York SAFE for Kids Act does, to turn them off by default unless adults opt in.

Examining each harm separately suggests the risk lies not in the platforms but in particular design choices. Most of these elements can be altered by the companies or, where necessary, age restricted, without removing young people from these services altogether.

A wholesale ban could then be explored if those targeted measures fail or platforms don't comply. Indonesia and Canada are now taking that approach by requiring that these platforms meet defined safety criteria before children can create accounts.

Learn from other industries

For features that do require age verification, we need to develop safety systems similar to those in areas like architecture, drug testing and food service. Such infrastructure can be considered "high-stakes," because it is implemented when laypeople cannot directly evaluate a product’s safety themselves and the consequences of failure are severe, such as the collapse of an apartment building.

We argue that online identity verification matches this pattern; no user can verify whether a vendor or platform properly handles and disposes of the photographs and identity documents required to verify their age, and the consequences of failure — misuse by the platforms or identity theft — are severe and often irreversible. For example, in 2025 the Federal Trade Commission received over 1.3 million reports of identity theft, with estimated losses of more than $15.8 billion.

We do not endorse specific technical standards, as our experience in counter-abuse technology shows that adversaries continually work to subvert online protections. Just as a bridge decays over time due to weather and usage, and needs to be maintained, regularly inspected and eventually replaced, we need to continue recognizing and adapting to new online threats over time.

Three attributes common to existing high-stakes safety systems would help address online age verification.

First, expert-led committees typically set and update safety standards, which agencies such as the U.S. Food and Drug Administration adopt and enforce. The technical standards behind age verification are very new, and they don't have the same level of government support to help develop or enforce them. For example, Google and Apple take different technical approaches to secure age verification in their mobile operating systems, which could make different systems less compatible over time. We strongly encourage governments to invest in supporting information security, internet and web standards bodies to move their work forward more quickly and to help the entire tech industry converge on common practices. Those standards should also be revised on a fixed schedule, in the same way that the building code is updated every three years.

Second, products in most high-stakes industries require review before they can be rolled out. Drug trials are extensive and scientifically rigorous, buildings require certified plans and inspections, and restaurants require a health certificate before they can serve a single meal. Age-verification vendors should face similar preapproval inspection, with computer architecture assessments, code reviews and adversarial security testing. Critically, social media companies should pass an equivalent review before deploying age verification.

Third, our food, drugs and buildings are safer than they otherwise would be because we fund full-time government experts in proportion to commercial activity in those sectors. This often takes the form of significant up-front fees for the companies involved ‪—‬ pharmaceutical companies must fund drug trials, and contractors must obtain building permits. This money covers the salaries and operating costs of trained inspectors. Vendors and social media companies should pay similar high fees, both up front for pre-review and annually for ongoing supervision, scaled to each company's audience size and scope of services.

In conclusion, we call on governments to approach child safety online as an ongoing challenge rather than a one-time regulatory fix. Making children safer online will take years of continued research into both child behavior and new technologies. One useful model is the progressive regulation of automobiles. Since the mid-1960s, new technologies and government targets have driven tailpipe emissions for most passenger vehicles down 99% and fatalities per mile down 78% per million vehicle miles traveled. If we take a similar approach to reducing online harms to our children, we are confident that we can achieve similar remarkable improvements in the years to come.

This article is for informational purposes only and is not meant to offer medical advice.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

For space fans and skywatchers, Wednesday, Aug. 12 is destined to be a date for the history books. Viewers in North America — including Alaska, eastern Canada and parts of the northeastern U.S. — will be able to see a partial solar eclipse during the daytime.

Across the pond, a total solar eclipse will be visible from eastern Greenland, western Iceland and northern Spain. (If you're planning to see this rare event in person, it's important to wear certified solar eclipse glasses during the partial phases of the eclipse ‪—‬ which means throughout the entire eclipse for everyone in North America. We have some solar eclipse gear suggestions that can help you watch this skywatching event from anywhere.)

Once the sun goes down, the night sky will offer its own treat: the Perseid meteor shower, which peaks overnight on Aug. 12-13. Thanks to a new moon, stargazers will have an even better chance of catching these lights streaking across the night sky, with the best hours to watch occurring from midnight to 2 a.m. local time. According to The Planetary Society, observers in dark-sky areas can hope to catch between 50 and 75 meteors per hour. The dark skies also make it a great night to try observing the Milky Way or the six planets that will appear together before dawn.

In a Live Science poll, we asked our readers whether they would be observing today's solar eclipse or tonight's Perseid meteor shower.

More than 240 readers responded to the poll, which was published Aug. 10. At the time of writing, the results showed that 51% of respondents will be watching both events, with 21% interested only in the Perseid meteor shower. About 20% of readers noted that they're not in a great place to watch the solar eclipse, with one commenter noting, "I'm in Georgia, not a viewing area, but I'll be looking online!" (Space agencies NASA and ESA will both be offering live streams of the eclipse from Spain, starting around 1 p.m. EDT).

What was perhaps the most surprising is that 2% of readers admitted they will not be watching either event but will instead stay inside and “scroll on Instagram.”

But some of those who are planning on witnessing these rare phenomena are making it an experience to remember. One commenter mentioned, "My brother is heading to Spain for the eclipse, meeting with a group of surfers, they'll surf into the eclipse."

Will you be enjoying today's skywatching events? Let us know in the comments below.

For space fans and skywatchers alike, Wednesday Aug. 12 is destined to be a date for the history books. For those in North America — including Alaska, eastern Canada and parts of the northeastern U.S. — viewers will be able to see a partial solar eclipse during the daytime. Across the pond, the total eclipse will be visible from eastern Greenland, western Iceland and northern Spain. (For those planning to see this rare event in person, it's important to wear the right solar eclipse glasses, especially if you're watching the partial phase of the eclipse; we have some gear suggestions that can help you watch this skywatching event from anywhere).

Once the sun goes down, the night skies offer their own treat: the Perseid meteor shower, which peaks overnight on Aug. 12-13. Thanks to a new moon, stargazers will have an even better chance of catching these streaking lights across the night sky, with the best hours to watch occurring from midnight to 2 a.m. local time. According to the Planetary Society, observers in dark sky areas can hope to catch between 50 and 75 meteors per hour. The dark skies also make it a great night to try and spot the Milky Way, or the six planets that will appear together before dawn.

With so much going on in the skies on Aug. 12, will you be watching? Take our poll and let us know what you think in the comments below.

There has been talk of "rescheduling" cannabis for years, a regulatory move that would significantly ease federal restrictions on the drug. Now, with a recent push from the White House to expedite the rescheduling process, we may see that change come to pass.

Moving cannabis to Schedule III — a less-restrictive classification for drugs — would represent a very important shift in how the federal government treats cannabis research, in that it would officially state that cannabis has medical use. This landmark decision would encourage some of the largest scientific investment in cannabis research to date and help build some of the strongest evidence of the value of cannabis-based therapies.

However, the move to Schedule III should be viewed as a transition rather than an endpoint.

How cannabis is regulated, today

Cannabis has existed in a complicated space in medicine, policy and public perception for decades.

More than 6 million Americans are registered to use cannabis through state medical programs. Clinicians participating in state medical cannabis programs generally "certify" and "recommend" specific cannabis products to certain patients, rather than write traditional prescriptions. Additionally, a handful of cannabis-derived and cannabinoid-based medications have received formal Food and Drug Administration (FDA) approval.

However, since the Controlled Substances Act was enacted in 1970, cannabis has been classified as a Schedule I substance. This category is associated with the highest level of restriction, including substances such as heroin, MDMA, and LSD.

Under federal law, Schedule I drugs are defined as having a high potential for abuse and no currently accepted medical use. Given some cannabis-derived drugs are FDA approved, it may seem contradictory that cannabis remains a Schedule I substance. The key is that federal drug scheduling, FDA drug approval and state cannabis laws are separate legal and regulatory systems.

Federal law classifies cannabis as a Schedule I drug. Separately, the FDA evaluates and approves individual cannabinoid medications for specific medical conditions. Independently, states can authorize cannabis for medical or adult use under their own laws, even though cannabis remains illegal under federal law.

This three-tiered regulatory landscape can be confusing for patients, clinicians and the general public alike.

Rescheduling would be a boon for research

For a few years, federal officials have recommended moving cannabis to Schedule III, which includes drugs with accepted medical uses that are subject to fewer federal restrictions. Ketamine and anabolic steroids, like testosterone, are examples. Compared with Schedule I substances, Schedule III drugs are considered to have a lower potential for abuse and can be prescribed for accepted medical uses under federal law.

Rescheduling cannabis would thus create a regulatory environment that's more favorable to research, which could support the development of future FDA-approved cannabis-based medicines. Researchers are already exploring cannabinoids as potential treatments for conditions including chronic pain, post-traumatic stress disorder, inflammatory bowel disease and sleep disorders, but the available evidence remains limited for many uses.

Evidence is limited, in part, because for decades researchers studying cannabis have faced regulatory hurdles that extend beyond those encountered with many other controlled substances.

Studies using cannabis often require additional federal, institutional and local approvals, compared to research with less-restricted drugs. They have extra requirements around securely storing and handling cannabis. They require detailed record-keeping to document how research materials are received, stored and used, as well as Drug Enforcement Administration registration for the principal investigator conducting Schedule I research.

Additionally, federal restrictions have historically limited researchers' access to cannabis products that reflect those people regularly use, such as commercially available flower, vape products and edible formulations.

This regulatory burden has historically discouraged some universities, healthcare systems and funding organizations from investing in cannabis research. These hurdles have made cannabis studies slower, more costly and smaller in scale than those conducted with more loosely controlled drugs.

Shifting cannabis to Schedule III would help to reduce these barriers, opening opportunities for more universities and healthcare systems to conduct much-needed clinical studies for medical cannabis. It could help researchers conduct the larger, more rigorous clinical trials needed to determine which therapies are truly safe and effective — and which are not.

Such gold-standard studies would better evaluate cannabis' potential benefits and risks, as well its effects in specific medical conditions and appropriate dosing for each. This evidence could, in turn, help reduce the stigma associated with cannabis use.

What rescheduling won't accomplish

Rescheduling cannabis will not automatically make cannabis more accessible to patients or immediately establish where cannabis is beneficial, what doses are most appropriate or which patients are most likely to benefit.

It would not automatically convert existing state cannabis programs into the conventional prescription model used for most Schedule III drugs. And it would not immediately resolve the complexities surrounding state medical cannabis programs, which operate outside the traditional FDA-approved medication pathway.

Unlike most prescription drugs, medical cannabis is an umbrella term that includes hundreds of products that differ in cannabinoid content, formulation, dose and route of administration. Most of those products have not been approved by the FDA, so rescheduling would not automatically place them into the traditional prescription drug system or immediately resolve many of the financial and access barriers patients face when trying to use them.

Without robust follow-up research, rescheduling alone would do little to resolve the uncertainty surrounding many cannabis products.

In short, the greatest benefit of moving cannabis to Schedule III would not be the policy change itself; it would be the opportunity to build a stronger evidence base for cannabis science and medicine. Policymakers may open the door to Schedule III, but researchers, clinicians and regulatory agencies will determine whether this opportunity will actually lead to the appropriate and meaningful advances in cannabis science and medicine.

The value of rescheduling will be measured not by how quickly it changes patients' access to cannabis, but by whether it helps create the scientific foundation needed to guide responsible medical use.

This article is for informational purposes only and is not meant to offer medical advice.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

In the weeks around the 2026 annual meeting of the American Society of Clinical Oncology, my phone kept buzzing with alerts about GLP-1 drugs and cancer. The headlines were everywhere — from NPR and The Washington Post to Substack and heated exchanges on social media — all circling the same claim: Ozempic might lower the risk of cancer.

Behind those headlines is a real wave of studies, involving millions of patients. I'm a physician and clinical epidemiologist, and my team and I design and interpret these same kinds of studies that test what widely used drugs actually do.

The excitement around the idea that GLP-1 drugs can prevent cancer is running well ahead of the evidence. Not wrong, necessarily. Just not yet earned.

Cancer is among the hardest things to study, because it is not one disease but more than a hundred. Breast cancer, lung cancer and blood cancers are not variations of the same disease; each has its own distinct biology and its own mix of genetic, environmental and behavioral risks.

A single drug never gets one verdict on how it affects cancer. It gets a hundred, raising the risk of some, lowering it for others and leaving most untouched.

The early cancer concerns

Before the current buzz that GLP-1 drugs could reduce cancer risk, the worry ran in the opposite direction: that they might raise cancer risk.

Researchers were originally concerned about thyroid cancer. Studies in rodents found that Ozempic caused thyroid C cell tumors, which is why U.S. regulators added a black box warning in June 2026 to the drug that advises against its use among those with a personal or family history of related conditions.

But rodents are not people. Human thyroid C cells are less sensitive to GLP-1 drugs compared to rodent C cells because they have far fewer GLP-1 receptors. Long-term studies in monkeys did not show the same abnormal thyroid cell growth seen in rodents.

A 2025 analysis of data from 93 clinical trials found no clear link between taking certain GLP-1 drugs and thyroid cancer. European regulators reached the same conclusion in 2023 after reviewing available research on GLP-1s.

Pancreatic cancer was another concern. Here, too, researchers found no consistent increased risk from taking GLP-1 drugs in a 2025 analysis of data from 62 studies.

The reassurance is real but provisional. These drugs are young, and cancer can take decades to surface.

The cancer story turns on its head

GLP-1 drugs, once scrutinized for possibly causing cancer, are now being discussed as drugs that might prevent or even treat cancer.

A 2024 study of more than 1.6 million people with Type 2 diabetes found lower rates of 10 out of 13 obesity-related cancers in people treated with a GLP-1 drug compared to insulin.

A 2025 study of about 87,000 adults found that overall cancer rates among those who started a GLP-1 drug were roughly 17% lower, with the clearest reductions in endometrial and ovarian cancers as well as a type of brain cancer called meningioma. The risk of kidney cancer was 38% higher among patients taking GLP-1 drugs, though larger studies are needed to confirm its significance.

A 2026 study of more than 110,000 women undergoing breast imaging found a roughly 30% lower risk of breast cancer among those who use GLP-1 drugs compared to those who do not.

Researchers are also studying how GLP-1 drugs affect a patient's chances of surviving cancer. Among more than 6,800 colon cancer patients in a 2024 study, around 16% of the 103 on a GLP-1 drug died within five years compared to 37% of those who were not taking one. A study presented at the 2026 American Society of Clinical Oncology reported a 34% lower risk of death across six cancer types for those taking GLP-1 drugs.

Why the evidence is so easy to misread

While these findings are exciting, studies on the effects of GLP-1 drugs on cancer risk have three features that make them especially easy to misread.

The first is healthy user bias. People who start a GLP-1 drug tend to be healthier and wealthier than those who do not. A person with obesity who has insurance, sees doctors regularly and can afford Ozempic is far more likely to start the drug than someone with the same height and weight who lacks those advantages.

Healthy user bias is very difficult to scrub out of observational studies, which compare people as they are treated in real life rather than assigning the drug at random. The very advantages that made it possible to receive a GLP-1 prescription, not the drug itself, may be what lowered a patient's cancer risk.

The second is the choice of the comparison drug. To understand the effect of a drug, it has to be measured against something else. What it's measured against will shape the results of the study. The 2024 study reported that patients taking GLP-1 drugs saw large reductions in cancer risk compared to those taking insulin. But comparing patients on GLP-1 drugs with patients taking the diabetes drug metformin showed no clear reduction in cancer risk.

Insulin is reserved for patients with more advanced diabetes, a condition that is itself a risk factor for cancer. In these studies, patients on insulin start out with a higher risk of cancer. Measured against a group at higher risk of developing cancer, almost anything can look protective. The apparent benefit came from the comparison drug, not the GLP-1 drug.

The third is timing. It is not so much another source of bias as a clue that other biases may be at work. Cancer can take decades to develop, yet most GLP-1 studies follow people for only a handful of years. In some studies, the apparent cancer benefits of GLP-1 drugs show almost immediately. But prevention doesn't work that quickly. A drug that truly lowered the risk of developing cancer would show its effect gradually as fewer tumors surface over time, not within the first few months of taking a drug.

When cancer risk appears to drop almost as soon as a patient starts treatment, the speed is not a magical triumph of GLP-1 drugs. Rather, in my view, it is the tell: People who start these drugs were already at lower cancer risk, and the GLP-1 drugs are getting credit for that preexisting advantage.

Moreover, almost all of this research comes from a handful of high-income countries, even as uptake of these drugs is increasing globally and the burden of cancer has disproportionately grown in lower-income countries. The idea that GLP-1s may reduce cancer risk for much of the world is being inferred from data generated almost entirely in a few rich ones.

What do randomized trials show?

The cleanest way around these study biases is randomized trials, which by design make comparison groups alike from the start.

Available clinical trials tell a quieter story than the headlines. Two 2025 meta-analyses, which pool data from multiple studies — one covering 50 trials and a second covering 48 — found little evidence that GLP-1 drugs either raise or lower cancer risk.

The findings of these meta-analyses are generally more reliable than those of observational studies because their total amount of data increases their statistical power. Despite their strengths, these analyses inherit the limitations of the trials they include. Most of the trials in these 2025 studies had a short follow-up of a year or two and recorded too few cancer cases to settle the question.

Randomized clinical trials designed to answer the question of whether GLP-1 drugs affect cancer risk would need to enroll tens of thousands of people and follow them for many years. Short of that, the next best evidence will come from observational studies designed to emulate randomized trials.

Until then, the idea that GLP-1 drugs lower cancer risk is a hypothesis worth testing but not a conclusion to act on.

The bottom line

Based on the available evidence, the clearest conclusion is the reassuring one: GLP-1 drugs do not appear to raise overall cancer risk. However, more ambitious claims that GLP-1s actively prevent cancer or improve survival after diagnosis remain unproven.

Suppose GLP-1s do have an effect on cancer risk. Researchers still would not know where that effect comes from: weight loss itself, broader improvement in metabolic function, or a more direct effect on inflammation, the immune system or tumors.

The surest thing I can say is also the least satisfying: It's still too early. These drugs are far younger than the cancers they are being credited with preventing.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Female farm managers are hidden in plain sight in ancient Roman texts, mentioned in laws, literature and grave inscriptions across five centuries. Modern historians have generally assumed they were housekeepers, in charge of domestic tasks and household meals, and segregated from the productive business of the farm.

My new paper, published in the Journal of Roman Archaeology, challenges this assumption.

In fact, there is evidence to suggest that many Roman female farm managers supervised wine production and other processes vital to farming and profits.

A false lead

A farming manual written by the Roman writer Lucius Junius Moderatus Columella in the 1st century CE is a window to the roles of the female manager.

An upper class landowner himself, Columella lists the responsibilities of farm managers, who were probably enslaved. The male manager was termed the vilicus, and the female manager the vilica, terms derived from their roles at the "villa".

But many historians reading this text have been side-tracked by a false lead: Columella begins his section on the vilica with a long quote from the Greek philosopher Xenophon, who wrote in Athens more than four centuries earlier.

The "natural" role of women, according to Xenophon, was to work indoors. Therefore in his discussion, the ideal upper-class married woman is depicted staying inside her town house to supervise the domestic work of the slaves.

The Greek writer is not discussing farm workers. In fact, Columella says four times that these ideas are not his own, but those of Xenophon, writing long ago.

Columella lists a completely different set of responsibilities for the vilica on a Roman farm: most importantly, the making of wine and olive oil, which were highly valuable commodities — the backbone of landowners' profits.

In his description of the vilica's duties, Columella includes the extraction of the juice from grapes during the harvest; adding flavorings and preservatives such as salt, wormwood, fennel or boiled grape-juice; and overseeing successful fermentation into wine.

According to Columella, the vilica also managed the processing of other farm products to make them long-lasting and profitable, such as turning inedible olives into olive oil for sale.

From archaeology we know that production of wine or oil, using huge machines in substantial buildings, could reach 50,000 to 100,000 litres per year, or even more. The vilica was therefore responsible for overseeing large-scale work essential to the operation of the estate.

Calling on the gods

Correct sacrifices to the gods were seen as vital to the success of the farm. Archaeology has uncovered altars for offerings in Roman wine-making buildings.

Wine-making in the Roman world was a precarious process, due to uncontrolled temperatures, bacteria or oxygenation. Wine could easily go moldy or turn into vinegar.

A striking element of Columella's account is that he includes carrying out offerings to avert such a disaster in his instructions for the female farm manager.

Looking for more clues

Other texts contain barely more than a mention of the vilica, revealing only her presence.

But we can still put together some clues.

Legal writings on inheritance, quoting the 1st century BCE jurist Trebatius, include her in the instrumentum fundi — whatever (including enslaved personnel) is required for productive work, gathering and preserving produce of the estate.

Another aristocratic landowner and writer, Cato the Elder, living two centuries before Columella, lists both the female and the male farm manager as essential staff for a vineyard or olive farm.

He devotes only one small section to outlining the female farm manager's tasks, but a close look shows these are not predominantly domestic. Cato includes keeping poultry and processing of seasonal farm products. Although he does include supervision of cleaning, this could refer to maintenance of work spaces such as stables and wine making buildings, an integral part of estate management – and also listed by Columella among the vilica's duties.

Cato also gives the vilica the responsibility for making sacrifices to the gods for the success of the farm. She must regularly offer garlands at the altar "for abundance", he writes.

A Roman mosaic showing estate work in different seasons depicts a scene of sacrifices for abundant crops to the god Jupiter (in his Celtic form as a sky and weather god). A woman is shown holding an offering of garlands, just as Cato instructs. Beside her are a jug for wine offerings and a male figure.

Perhaps this mosaic shows the vilica and vilicus, both essential to the success of annual harvests. While such images of female figures are rare, one fragmentary wall painting from Rome shows a female overseeing wine-making workers, just as Columella describes.

This combined body of evidence suggests we should expand outdated understandings of the roles that Roman women played in farm production, the dominant sector of ancient economies.

No vilica has left us an account of her work in her own words. But by paying careful attention to the evidence, we can hear an echo of her voice.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

From Augustus to Nero, see how much you know about ancient Rome's famous leaders with our Roman emperor quiz!

On July 14, 2026, "Gus," one of the most complete specimens of Tyrannosaurus rex, went to an as yet unidentified buyer for US$50.1 million. This auction at Sotheby's set a record for most valuable fossil ever sold. Another dinosaur has entered the luxury collectibles market, a reminder that even Earth's deepest history can be sold to the highest bidder.

To paleontologists like me, however, a fossil like "Gus" — excavated from the Hell Creek Formation in South Dakota over three years starting in 2021 by commercial collector Thomas Heitkamp and his team — is not a trophy or a work of art. It is an irreplaceable scientific archive. Fossils preserve evidence of evolution, extinction, growth, disease, injury and ancient ecosystems. They are finite, nonsubstitutable records of life’s history on Earth.

Science depends on independent verification of claims and healthy debate. Researchers must be able to revisit specimens, test earlier conclusions and ask new questions.

But once a scientifically important fossil enters a private collection, access for researchers is no longer guaranteed. Collectors typically sequester their fossils in their homes. Even when privately owned specimens are loaned to museums, the owners can change their minds, ending access at any time. This issue is especially of note when it comes to Tyrannosaurus rex; a 2025 study found that while there were 61 T. rex fossils in public trusts at that time, 71 were privately held.

That is why the Society of Vertebrate Paleontology, of which I'm a long-term member and president-elect, has long argued that scientifically significant vertebrate fossils belong in the public trust, curated in museums and universities that preserve them permanently, make them available for research and share them with the public.

Finding a fossil

Supporters of commercial fossil sales often argue that without sales to private collectors, specimens like "Gus" would remain buried or erode away. They’re right about one thing: Discovery matters. Many extraordinary fossils have been found by ranchers, hikers, amateur collectors and commercial excavators. Paleontology is accessible to everyone who has an eye for observing nature — you don't need to be an expert with academic credentials to make an important discovery.

But discovery is only the beginning. A fossil's scientific value depends on careful documentation of where it was found, the rocks surrounding it, and the plants and animals preserved alongside it. Those details allow scientists to reconstruct ancient ecosystems, understand how an animal lived and died, and interpret how its remains became fossilized. When that contextual information is incomplete or lost, much of the fossil's scientific value is lost as well.

Yet even discovery, excavation and publication barely scratch the surface of a fossil's scientific importance. The greatest scientific value of a specimen often comes decades later, when researchers ask new questions and apply new technologies that earlier generations never imagined. A specimen that seems fully studied today may yield surprising new information tomorrow, but only if it is still available for study.

Delayed discoveries

Consider the iconic dinosaurs, including T. rex, Triceratops, Diplodocus and Stegosaurus, first collected more than a century ago. Early paleontologists could describe their shapes but had no way to dig deeper by peering inside the bones. Because those specimens were preserved in museum collections, later generations could revisit them with technologies that didn't exist when they were discovered.

Paleontologist Larry Witmer and his collaborators at Ohio University started using CT imaging 20 years ago to reconstruct the internal anatomy of historic dinosaur fossils without damaging them, based on how X-rays travel through specimens. Brain cavities, inner ears, air spaces, nerves and blood vessels became visible for the first time, revealing how dinosaurs balanced, heard, smelled and perceived their world.

Henry Fricke, Thomas Cullen and other geochemists have used isotopic signatures preserved in fossil teeth and eggshells to reconstruct dinosaur diets, migration patterns and body temperatures. This research has revealed how dinosaurs lived: what they ate, how they moved through ancient landscapes, and even how warm their bodies were.

More recently, molecular paleontologist Jasmina Wiemann and her collaborators have identified chemical traces preserved in fossil bone, eggshell and skin that reveal aspects of dinosaur biology unimaginable even a generation ago. Until now, paleontologists had no way to know details about metabolic rates and reproduction or the colors of skin, feathers and eggs.

In my own research I use microscopes to uncover the hidden stories preserved inside dinosaur bones and teeth. Thin sections of fossil bones reveal that dinosaurs grew more like mammals and birds than like oversized reptiles. Microscopic modifications to bones capture traces of ancient scavenging, and tiny signatures deep inside baby dinosaur bones indicate the moment of hatching.

None of these discoveries would have been possible if the original fossils had vanished into inaccessible private collections.

Shared natural heritage, on the auction block

Fossils are not static objects whose scientific value is exhausted once they are described. Their value grows as science advances, but only if future researchers can continue to examine the original specimens.

Of course, sometimes dinosaur fossils are rescued from obscurity through purchase and immediate deposition or donation to natural history museums. Some of the world's most important dinosaur fossils are accessible today because individuals, companies or organizations with the means to acquire extraordinary specimens recognized that they belong where scientists can continue to study them and where future generations can learn from them.

Purchasing a fossil in order to place it permanently in the public trust is fundamentally different from acquiring it as a private collectible: One expands access, the other leaves access uncertain.

But as fossil prices rise into the millions, museums increasingly cannot compete. The most significant fossils are no longer reliably entering public collections. Instead, they are becoming luxury assets whose market value supersedes their scientific value.

Dinosaurs belong to our shared natural heritage. They inspire wonder because they connect all of us to a world unimaginably older than our own. For me, the question raised by auctions like the one on July 14 of "Gus" is not who can afford to own these relics of the past. It is whether future generations have the chance to study and learn from them.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

How much do you know about the king of the dinosaurs? Test your knowledge with our T. rex quiz!

The use of nicotine pouches is increasing among teenagers and young adults in the U.S. As researchers who study nicotine use among young people and develop strategies to prevent it, we worry these discreet products might be driving a new public health crisis. It may be that more and more teens are becoming addicted to nicotine pouches under our watch.

To stop that crisis in its tracks, we need to act now. Myriad strategies would help, including eliminating flavored pouches and reducing the amount of nicotine in the products, as well as regulating their marketing and stopping companies from making deceptive claims. And while we await such regulation, informing educators, parents and teens about the harms of nicotine pouches will be key.

Why worry now? Nicotine pouches — small, pillow-like products that users place between their gum and lip — are now the second most used tobacco product among U.S. adolescents. In a recent survey, 2.3% of high school and about 1% of middle school students, an estimated 460,000 students total, reported using a pouch in the past month. More than 9 in 10 pouch users opted for flavored pouches, with mint being the most popular.

While rates of pouch use remain behind e-cigarette use, more students reported using pouches than cigarettes, cigars, smokeless tobacco or any other form of nicotine. Nicotine pouches were the only product whose use increased in the survey, while the popularity of other products declined or remained stable.

What makes nicotine pouches particularly tricky is how invisible they are to the outside observer. Unlike cigarettes, they produce no smoke. Unlike e-cigs, they generate no visible aerosol. They are small and easy to conceal. That invisibility is not a coincidence; it's part of big tobacco companies' intentional design. An adolescent can use nicotine pouches during class and no one around them would know.

Each pouch contains nicotine powder that is either derived from tobacco plants or synthetically produced, as well as other ingredients like acidity regulators and flavorings. Those flavors can be dessert-like, fruity, minty, and reminiscent of coffee or cocktails — all of which appeal to young people.

The amount of nicotine per pouch varies by brand, but generally, using one pouch is roughly equivalent to smoking one to four — or more — cigarettes. That's a problem because, although pouches don't produce smoke, nicotine is still very addictive and harmful to the developing brain.

Nicotine use can negatively affect teens' attention and impede their impulse control, hurting their academic performance and mental health. In regards to physical health, data suggest that smokeless oral nicotine products may harm some aspects of heart health and raise the risk of death in people with existing heart disease.

It may be tempting to wave away concern because the percentage of adolescents using these pouches appears relatively small. However, that would be a mistake, as doing so would mean watching history repeat itself.

Rates of e-cigarette use also started low, and by the time national data finally showed a clear increase, their use was already a deeply entrenched behavior among adolescents, with many addicted. Furthermore, national surveillance studies are, by design, snapshots in time and often lag behind what is happening in real time in classrooms and communities.

We're concerned that more teens are becoming addicted to nicotine pouches, now, and want to curb the trend. At Stanford Medicine's REACH lab, we work with thousands of schools and teachers across the U.S. every year, providing free, evidence-based tobacco prevention curriculums. We've spoken with hundreds of educators, parents and students across the country who have raised significant concerns about nicotine pouches, and presented talks to over 1,200 people on the subject.

It became clear that we must create a specific approach to prevent teens from using pouches — so, we created a course called Not So Sweet: Oral Nicotine and Smokeless Tobacco Prevention Curriculum.

The free course is designed to be used in schools or by anyone who works with teens. It explains the different types of oral nicotine and smokeless tobacco products and their many health effects, including addiction. It also lays out the marketing tactics employed by the tobacco industry to attract youth in the first place. In just a few short weeks of releasing the curriculum, we already have over 1,500 active users.

We're heartened that people are already engaging with the course, because we might already have a public health crisis on our hands.

Today's nicotine pouch market in the U.S. is dominated by three brands: the top seller, ZYN; On!; and VELO. These brands are owned by Philip Morris International, Altria, and British American Tobacco, respectively. The companies have decades of knowledge in marketing nicotine products that they are now deploying to sell nicotine pouches.

They're falling back on successful advertising tropes, telling consumers that nicotine pouches could help them socialize, unwind, or appear cool to potential romantic partners. Ads also emphasize the discreet nature of the pouches, which allows them to be used anywhere without drawing too much attention.

Through these marketing strategies, nicotine pouches have soared in popularity and sales, Zyn has become a household name, gaining nicknames like Zynnie, Zynfluence and Zyndulgence. Zyn has effectively become the "Coke" of colas, the Kleenex of tissues, the Juul of e-cigs.

What's more, the Food and Drug Administration just authorized various Zyn products to be marketed with the claim that using Zyn instead of cigarettes puts you at a lower risk of mouth cancer, heart disease, lung cancer, stroke, emphysema and chronic bronchitis. But when it comes to teenagers, we see many using Zyn or other nicotine pouches alone or alongside other nicotine products that aren't cigarettes. Unlike adult smokers who may be switching to pouches, teens may be newly introduced to nicotine addiction through products like Zyn.

It is not a matter of cigarettes versus Zyn for many teens; it is Zyn versus nothing.

Nicotine pouches might come in tantalizing flavors, but they are not sweet — they are dangerous, especially for adolescents. Reminiscent of what we've seen for decades with cigarettes and over the last decade with e-cigarettes, we once again have a nicotine product that is very appealing to and being increasingly used by young people. Yet, they contain a great deal of nicotine and are unsafe. We need to work together, now, to ensure we don't perpetuate another generation of addicted kids.

Editor's note: Bonnie Halpern-Felsher has served as an expert witness in litigation against various tobacco companies.

This article is for informational purposes only and is not meant to offer medical advice.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Consider the following scenario. Suzy is 63, recently retired, and trying to decide when to start receiving Social Security and how to manage her retirement savings to minimize the tax hit.

She opens an AI chatbot, types in the details and gets a calm, well-organized and confident answer: Claim now, convert this much, here is the reasoning.

The chatbot sounds authoritative and even shows its work. So Suzy follows its guidance and never calls a financial planner. Maybe the advice was fine. But maybe it quietly ignored the fact that Suzy's spouse is younger and in poor health, which can flip the Social Security math. It also may have overlooked that the retirement savings plan conversion it suggested would push Suzy into paying higher Medicare premiums two years later.

Suzy won't find out for a long time, if ever, whether this guidance was right for her. And the AI will never call back to say it was unsure.

Suzy isn't an exception. AI chatbots have entered everyday life with remarkable speed: A 2025 Pew Research Center survey found that 34% of U.S. adults and 58% of those under 30 have used ChatGPT, roughly double the share two years earlier.

A growing number are asking AI about money, and some are getting burned. According to a 2025 survey of 2,000 U.S. adults by Pearl.com, a professional services platform, 19% said they lost more than $100 by following financial advice from an AI chatbot. Among Gen Z investors, that figure rose to 27%.

These aren't hypothetical risks. People are already paying for answers about their money that are confident — and wrong.

As a finance professor who has been closely watching the spread of AI into personal finance, this is the part of the AI story that worries me most. And it's not the part you usually hear about.

We argue about AI the wrong way

There are two seemingly opposite complaints about AI. One is that people trust it too much, treating a chatbot like an oracle, a tendency researchers call algorithm appreciation. The other is that people don't trust it enough and dismiss its useful tools, a tendency known as algorithm aversion.

I argue these are actually two sides of the same coin, and what decides which side you see is whether you can tell when the AI is wrong.

When an AI fails in an obvious way, you notice and lose confidence. So you're more likely to seek a professional or another human you trust sooner than you otherwise would. That is the safe failure.

The dangerous failure is the opposite. The answer is fluent, confident — and wrong. You have no way to catch it, so you keep managing the problem yourself long past when you should have asked for help.

The trouble is that with money, the second kind of failure is the common kind.

When you mistake fluency for accuracy

Three things make financial advice especially treacherous for AI.

First, fluency is not accuracy. People naturally read a confident and well-articulated answer as competent. But how polished an answer sounds tells you almost nothing about whether it fits your situation or the accuracy of the proposed solution. A chatbot can be word-perfect and still be wrong about your taxes, because your taxes depend on details it never asked about.

Second, AI is least reliable exactly where the stakes are highest. AI tools are good at routine and general topics: what a Roth IRA is, how compound interest works, the difference between a stock and a bond.

But financial life is full of rare, complicated, one-time decisions: exercising stock options, understanding the alternative minimum tax, making required, minimum 401(k) distributions, deciding on a Social Security strategy as a couple, drawing up a divorce settlement.

I made a similar argument three years ago about AI trading on Wall Street. Because market crashes are rare, there's little data for AI to learn from, so it can be most confident exactly where it is least informed.

That worry hasn't faded. Market watchers now caution that AI trading bots are creating fresh financial risks, and that same blind spot applies to your personal finances. Researchers call this uneven competence a "jagged frontier" — reliable with common cases but unreliable for unusual ones. And in finance, the unusual cases tend to be the expensive ones.

Third, you often can't check the work. Financial advice is what economists call a "credence good," like a mechanic's diagnosis or a doctor's recommendation. You often can't tell whether the advice was good, sometimes for years. A mistaken tax move may not surface until an audit. A bad 401(k) drawdown plan may not bite until the stock market slumps. Without quick feedback, the wrong-but-confident answer never gets corrected.

This is why the Pearl numbers above are probably an undercount, since they capture only losses people noticed.

The quiet failure is the one to watch

Notice that the real harm in Suzy's story isn't a single dramatic mistake. It's that a confident answer made Suzy feel no need to call a professional, so the call never happened.

The danger is not so much that you act on bad advice but that you never seek good advice. The smoother and more reassuring the tool, the easier it is to stay in do-it-yourself mode past the point when you need outside help.

Who's most at risk? In a study of a large robo-advising platform in India, co-author Vishaal Baulkaran and I found that its users skew young, are predominantly male and tend to be smaller retail investors and professionals. And new account sign-ups rise during periods of high market volatility.

In other words, the people leaning hardest on automated advice match that 27% figure among those Gen Zers who lost more than $100 while using a chatbot for financial advice. They reach for it just when markets turn turbulent and a wrong move is most costly.

There's also an incentive worth naming. In my new analysis, I argue that a tool that earns its revenue by holding your attention has a reason to sound confident and helpful: Confidence keeps you on the platform. The catch is that the user it retains that way is sometimes the one who should have been handed off to a human.

A system tuned to keep you engaged isn't the same as one tuned to protect your financial future, and the two can point in different directions. The disruption is already underway, as wealth managers face what Bloomberg has called a chatbot reckoning. A single, new AI tax tool recently sent wealth management stocks sliding as investors bet that automated advice will eat into the business.

How to be smart about using AI

These findings don't mean that people should avoid AI for money advice. Used well, these tools are a valuable and free financial educator.

This is also not to say that a financial adviser always has the right answers. As with finding any kind of specialist, it's important to do research first and make sure they meet the kind of criteria laid out by the Consumer Financial Protection Bureau. Fee transparency is also crucial.

But if you do turn to AI, the skill is knowing where to draw the line.

Treat AI as a starting point, not a verdict. It's excellent for learning concepts, drafting questions and getting oriented before a meeting. It can teach people the vocabulary to have a smarter conversation with an expert.

But watch out for the signals that you have left its comfort zone and entered the territory where AI is weakest and a confident answer is least trustworthy. The red flags are large dollar amounts, tax consequences, anything irreversible and anything that turns on the specifics of your situation rather than a general rule.

Estate questions, the drawdown of retirement savings, strategies for claiming Social Security benefits, business structure and major one-time transactions all belong in this category. Those are the decisions that call for bringing in a human, such as a certified financial planner.

And remember, confidence isn't competence. When the answer about your money sounds most polished and most certain, that's not a reason to relax. On the hardest questions, that smooth confidence is exactly the signal that you should pick up the phone and talk to an expert.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Tropical forests draw down and store large quantities of CO₂ from the atmosphere. The Amazon rainforest in South America, for example, stores approximately 123 billion tonnes of carbon — more than is stored in any other terrestrial ecosystem in the world. But these forests are facing a critical challenge.

Research from 2023, which was carried out by me and more than 100 colleagues, found that tropical forests in South America are vulnerable to climate extremes. We determined that during an El Niño event, the warm phase of a natural fluctuation in the Earth's climate system, South American tropical forests may fail to act as a carbon sink.

This finding becomes even more alarming when we consider the increasing frequency and intensity of El Niño events. There have been twice as many "very strong" El Niños in the last 60 years as there were in the 60 years before that. And the US National Oceanic and Atmospheric Administration has recently confirmed that such an El Niño is currently underway.

Tropical forests absorb CO₂ through the process of photosynthesis and convert it into biomass. However, the balance between photosynthesis and respiration is delicate and depends on two factors: temperature and water availability.

In hotter and drier conditions, plants close the pores of their leaves to avoid water loss. But closing them effectively cuts off a plant's fuel supply because it is through these pores they absorb CO₂. This starves plants of the carbon needed for photosynthesis and to grow.

During El Niño years, which are characterized by high temperature anomalies, prolonged climate stress leads to reduced forest growth and increased tree mortality. The effects of this are felt for decades as carbon is released back into the atmosphere when the dead trees decompose.

Our findings revealed that during the 2015-2016 El Niño, when temperatures on land were at least a degree higher on average than usual conditions, some of South America's tropical forests effectively stopped absorbing carbon. This raises concerns about the possible impact of the current El Niño on the Amazon and global climate.

In our research, we measured over half a million trees across six South American countries over a period of more than 30 years, using tape measures to track their growth. These trees belonged to over 4,000 different species. We used this data to calculate precise estimates of the amount of carbon stored as a forest's aboveground biomass.

We found that the vulnerability of these forests to El Niño conditions was closely linked to their baseline climate. While we tend to assume that rainforests are all hot, wet and biodiverse ecosystems, seasonal drought is a reality for many tropical forests. Conditions in regions at the edge of the Amazon rainforest, for example, tend to be particularly hot and dry.

Our findings revealed that drier forests at the edge of the Amazon, where trees regularly endure periods of limited water availability, were especially susceptible to extreme El Niño conditions. On average, a 0.5°C increase in temperature caused these forests to lose 0.5% of their aboveground carbon.

Larger trees suffered the most. While tree mortality rates increased from 1.8% to 3% per year during the El Niño in South American tropical forests as a whole, mortality rates effectively doubled for medium (classified as over 20cm in diameter) and large trees.

The fact that larger trees with less dense wood died at much higher rates compared to small trees and those with high wood density points strongly to hydraulic failure, when intense atmospheric moisture demand snaps the tension in the tree's internal water column rather than slow carbon starvation.

These results suggest that adaption to seasonal drought may not be sufficient to protect tropical forests from extreme events. Climate extremes are possibly already pushing forests at the edges of the Amazon beyond their capacity to adapt, causing catastrophic carbon losses.

A looming threat

Scientists have warned that 2026 may again be the warmest year on record. Heightening the alarm further is the severity of the current El Niño. Never before has an an El Niño begun when oceans are already so warm and air temperatures so high.

On top of this is the fact that, over the past three decades, the edges of the Amazon have experienced some of the highest temperatures and most rapid warming the tropics have ever seen. The structural integrity of a forest is compromised when a major climate anomaly occurs before it has recovered from recent, multi-year stress.

These compounding factors mean that we risk witnessing tree and carbon losses on scales not yet seen.

Tropical forests are invaluable assets in the fight against climate change. But South American tropical forests, a once-reliable carbon sink, are vulnerable to intensifying heat and drought. There is a risk these essential ecological allies stop acting as a carbon sink as extreme climate conditions become the norm.

Preserving tropical forests is thus essential. Their ability to continue acting as carbon sinks hinges on efforts to protect them and a collective commitment to limit global temperature rise. The Amazon's future depends on this, and so does ours.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Rainforest quiz: Can you sort Earth's largest rainforests from biggest to smallest?

For decades, cities have been designed around "gray infrastructure." Roads move people. Bridges connect communities. Water systems protect public health. These systems are governed by engineering standards because society recognizes that safety cannot depend on good intentions alone.

Urban nature deserves the same recognition.

In June, a heat wave gripped Europe, breaking temperature records across the continent. In France, officials reported over 2,000 excess deaths. In the U.K., hospitals declared critical incidents and machinery and IT systems failed. In the U.S., a heat dome over the Midwest and East Coast disrupted Fourth of July celebrations, with at least 25 heat-related deaths over Independence Day weekend.

Cities are being tested by a warming climate. Yet one of our most effective forms of climate infrastructure is being lost — not because it does not work but because we still do not treat it as infrastructure.

My view is simple: Urban nature ‪—‬ including street trees, parks, wetlands and other urban green spaces ‪—‬ should be categorized and managed as essential infrastructure, with minimum standards for its protection, quality and long-term maintenance. Just as we regulate roads, bridges and drinking water, we need standards that ensure all urban residents benefit from healthy and thriving urban nature. Without such standards, cities will become hotter, less resilient and increasingly unequal as climate change accelerates.

Scientific research shows that urban trees and green spaces cool cities during heat waves, reduce flooding by absorbing stormwater, improve air quality, store carbon, support biodiversity, and improve both physical and mental health. During extreme heat, neighborhoods with mature tree canopies can be several degrees cooler than nearby streets dominated by concrete and asphalt. Those few degrees can mean the difference between manageable discomfort and dangerous heat exposure, particularly for older adults, children and people with existing health conditions.

But while the science is strong, our governance is not.

Cities enthusiastically announce ambitious tree-planting campaigns, biodiversity strategies and new greening targets. These initiatives are valuable, but they often focus on what is easy to count rather than what truly matters. Planting a tree is not the same as growing a healthy urban forest. Creating a park does not guarantee biodiversity. A green roof delivers little value if it fails during drought.

The real measure of success is whether urban nature continues providing benefits decades after trees are planted and established.

This is where cities are still falling short.

Unlike buildings or transport systems, urban nature rarely operates under consistent minimum standards. Many cities have no requirements for minimum tree canopy cover, adequate rooting space, soil quality, biodiversity targets, long-term maintenance or even whether newly planted trees survive. As a result, access to nature depends heavily on where people live. Wealthier neighborhoods often enjoy mature tree canopy and high-quality parks, while disadvantaged communities experience hotter streets, fewer green spaces and greater exposure to climate risks.

That is not simply an environmental issue. It is a public health issue, a climate adaptation issue, and an issue of social equity.

The solution is not simply to plant more trees. It is to establish urban nature standards that recognize nature as essential infrastructure.

These standards would not prescribe the same solution for every city. Instead, they would establish minimum expectations based on scientific evidence. They could include targets for accessible green space, minimum tree canopy cover, sufficient soil volume for healthy tree growth, biodiversity outcomes, long-term maintenance funding and routine monitoring to ensure that urban nature continues delivering benefits to communities.

Importantly, these standards should focus on outcomes rather than simple planting targets. Counting how many trees are planted tells us little about whether cities are becoming more resilient. Measuring survival, canopy development, ecosystem health and equitable access provides a much better picture of whether investments in urban nature are actually returning dividends.

Some critics argue that cities cannot afford stronger standards for urban nature. Municipal budgets are already stretched by housing, transport and aging infrastructure. But this argument overlooks a fundamental reality: We already spend enormous sums responding to the consequences of extreme heat, flooding, poor air quality and declining public health. Healthy urban ecosystems help reduce these costs while delivering multiple benefits simultaneously.

As climate change accelerates, the distinction between gray and green infrastructure becomes increasingly blurred. Trees cool buildings and reduce energy demand. Wetlands decrease flood damage. Parks improve public health and strengthen communities. These are not optional amenities. They are functioning components of urban infrastructure that deserve the same level of planning, investment and accountability as any engineered system.

Building codes transformed the safety of our cities because they established minimum standards that every development had to meet. Climate resilience now demands a similar transformation for urban nature.

The next generation of resilient cities will not be defined simply by how many trees they plant but by the standards they adopt to protect, restore and sustain the living infrastructure urban life depends on.

The real question is not whether cities can afford to invest in urban nature; it is whether they can afford not to.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

Artificial Intelligence (AI) is continuing to have a disruptive impact on ever more parts of humanity. But what does it mean in the long run? A new paper, available in pre-print on arXiv from Austrian researcher Sergey Ivliev, extrapolates what the wide scale adoption of AI means for the future of humanity in space — and in particular what it means for the ultimate question of whether we're truly alone in the galaxy or not.

A framework for much of the search for extraterrestrial intelligence came from famous physicist Enrico Fermi, who simply asked "Where is everybody?" at a lunchtime discussion at Los Alamos in the 1950s. Though never officially published, Fermi's lunch partners from that day have passed down an oral history of that conversation that has cemented it into the Search for Extraterrestrial Intelligence (SETI), at least until Michael Hart formally laid out the argument and mathematics for the underlying question in a paper in 1975.

There are plenty of potential answers to the Fermi Paradox, many of which can be found floating around the internet — and some are likely more valid than others. But Ivliev, a Ph.D. in Mathematical Economics and founder of environmental project consultancies such as Peatland Ecosystems and Vlinder, suggests a new resolution: the Quiet Expansion filter.

The central argument of the paper is that there aren't thousands of alien mega-structures lighting up the night sky because once a civilization reaches the threshold of Autonomous AI-Cosmoindustry (AICI), "loud", resource-hungry empires motivated by prestige or romance become irrational. However, this does not mean expansion stops; instead, it shifts to a "quiet" mode driven by rational goals like survival diversification, knowledge preservation, and scientific observation.

The AICI threshold is reached when a civilization possesses a self-sustaining off-planet industrial and computational system capable of designing, manufacturing, repairing and launching space hardware through AI-mediated autonomy. We're already taking tentative steps in this direction with the advent of space-based data centers, but true AICI —where a civilization can extend its infrastructure beyond its home planet without continuous biological intervention — is leaps and bounds beyond our current capabilities.

In this vein, Ivliev is drawing on work done by astrophysicists Sergey Popov, who noted that a truly rational AI system would reject human-like motivations for space travel — such as romance, conquest, or prestige. Instead, AI would view space expansion as simple risk management.

To an AI, putting all your eggs in one basket — whether that basket is a single planet, solar system, or even galaxy, can lead to a single point of failure. Therefore, expansion is highly logical as a way of mitigating the risk posed by that single point of failure. At the point where we have reached AICI, the cost for sending a 10kg interstellar probe to another star at 1% of the speed of light is roughly 4.5x10^13 Joules — a tiny fraction of the overall energy budget of such a civilization.

One key aspect is that the 10 kg probe doesn't contain any actual people — it simply holds the "seeds" to restart life elsewhere in case a catastrophe happens back in the home system. It would contain a civilization's knowledge, and possibly some of its biological material, enabling a sufficiently advanced AI to rebuild the entire civilization from scratch. This is the "Quiet Expansion" where an AI sends low-mass and hard to detect "seed systems" instead of moving millions of biological entities around in massive interstellar space ships.

There are some additional constraints on this method of expansion, including selecting promising exoplanets discovered by remote sensing and deploying minimal local resources to maintain themselves until needed. Additionally, the AI would restrict self-replication of the probes in order to avoid any "grey goo" scenario with a probe attempting to take over entire swathes of the galaxy.

This has obvious implications for why we've never found "loud" technosignatures. In this scenario, a "null" result of being unable to find the thermal signature of a Kardashev-III scale civilization doesn't mean a galaxy is empty. It just means that the successful civilizations are residing in a "quiet" state in case its back-up plans are needed.

But there's another, more ominous implication from this framework. If interstellar backups are cheap to make, and we haven't found any in our own backyard, that means either we're one of the first civilizations to make it to that point or the transition from a planetary industrial society to a space-based one is a narrow path to tread. Admittedly the probes such civilizations would send out are probably hard to find even in our own solar systems, but if we're unable to, it means we're ending uncharted territory — and might just run into a filter that had silenced the rest of the galaxy. That's a sobering thought, but one to keep in mind as we start to advance our own AI capabilities.

The original version of this article was published on Universe Today.

Are you a UFO fanatic? Find out with our extraterrestrials quiz!

During several recent conversations, people have told me that they've stopped checking their phones in the morning. Not because nothing was happening, but because everything was. They described the feeling as standing under a waterfall of perpetual bad news.

This experience is far from an isolated one. According to Reuters Institute's 2025 Digital News Report, 69% of Canadians at least occasionally avoid the news now.

Globally, 40% report they at least sometimes or often do the same, the highest figure ever recorded. People shared consistent reasons for this: the news put them in a bad mood, they felt overwhelmed and powerless to act.

As a researcher in developmental psychology, focusing on social development and psychological well-being, I argue that news fatigue is not laziness, weakness or a generational decline in civic interest. It's the predictable response of a human brain meeting an environment it was never designed to navigate.

Wired for bad news

Long before smartphones or even the printing press, our cognitive architecture was shaped by a single problem: stay alive long enough to reproduce. Our ancestors whose attention drifted past the rustle in the grass left fewer descendants than those who froze, looked and listened.

The brain that paid attention to threats was the brain that survived.

This is the foundation of what psychologists call the negativity bias, one of the most replicated findings in cognitive science. Across decades of research, the human mind has been shown to weigh negative information more heavily than positive, attend to it faster and remember it longer.

A predator nearby mattered more than a beautiful sunset. The cost of missing a real threat was death, while the cost of overreacting was a few minutes of wasted vigilance. The asymmetry made this bias adaptive.

Here is the problem: the human brain has not changed since then. We are the same species as we were thousands of years ago. What's changed is the size of the world it's asked to scan for threats.

Scanning the whole world

For most of human history, the threats our nervous system processed were local. A neighbouring tribe. A drought. The illness of a child we personally knew. Information about distant places would barely arrive, and if it did, it was mainly irrelevant.

In 2026, the same neurological system is being asked to absorb a war in one region, a financial shock in another, a climate disaster in a third and a violent crime in a fourth, all before lunchtime.

A study published in the scientific journal Nature Human Behaviour examined more than 105,000 real news headlines viewed nearly six million times. Each additional negative word increased click-through rates, while positive words had the opposite effect.

Recent studies suggest people around the world demonstrate measurably stronger physiological responses to negative news than to positive news. The body is reacting before the mind has decided whether the threat is relevant.

Some researchers have introduced a clinical framework for what happens in this instance called Problematic News Consumption (PNC) — a pattern of news engagement that results in preoccupation, dysregulation and disruption to daily functioning. In their 2022 study, the researchers found that 17% of American adults qualified as having severe levels of PNC. Among that group, 61% reported feeling unwell quite a bit or very much, compared with 6% of those who didn't.

For minority populations, news fatigue may be even more consequential.

Repeatedly witnessing harm directed at our own groups, even when we're not the immediate target, can have a significant psychological impact on people from the same group affiliation. For racialized communities, such as immigrants, the cognitive load could be even heavier, and the option to simply stop watching is much harder to exercise when the news is about their country of origin.

Looking away is not the fix

What's the solution to news fatigue? Well, it's not avoidance. A democracy depends on informed citizens.

Many adults already cite the spread of misleading information as a major source of stress. Withdrawing from accurate, trustworthy information only deepens the problem. We're wired to pay more attention to bad news, and that kind of content will find its way to us one way or another.

The fix is to manage the consumption and the sources.

Several approaches can help manage news fatigue and protect mental health. Containing news consumption to defined windows of time reduces the sense of being overwhelmed. Choosing depth over volume also matters: one carefully reported long-form article will inform you better than bursts of random, unreliable and emotionally loaded posts on Instagram.

There is also value in distinguishing between information and action — research on perceived control and stress consistently shows that the gap between awareness and agency is one of the strongest predictors of psychological distress. Identifying what you can actually do about what you read in the news, however small, regulates that response.

Finally, be wary of "rage bait" — intentionally provocative messages or content designed to boost engagement on social media platforms by eliciting negative reactions. Recognizing that certain content creators want to provoke rather than reflect reality creates useful cognitive distance.

The news will not become less "heavy." But our relationship with it can become more deliberate. Our brains were not built for this scale of input. They were, however, built to learn to adapt.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

As I sip coffee in my Berlin apartment and fire a question at Google's AI chatbot Gemini, it's easy not to think about the energy it takes to generate a response. Once the signal reaches my router, it whizzes, I assume, through copper wires or fiber-optic cables to one of Google's data center hubs. Somewhere inside the data center's labyrinthine halls of stacked processors, my query gets converted into numbers and undergoes billions of computations to determine context and meaning. The answer, once assembled, races back, in the blink of an eye.

Data centers — the beating hearts of the internet, powering everything from email to web searches — have existed for decades, but with the growing popularity of AI to generate text, images and video, they're using more energy than ever. According to Google's own estimates, processing a median-length text prompt with its AI assistant Gemini consumes around 0.24 watt-hours.

These amounts, individually small — 0.24 watt-hours is equivalent to watching TV for about nine seconds — are adding up fast. In March 2026, OpenAI estimated that more than 900 million people use its AI chatbot, ChatGPT, every week, tallying billions of queries daily.

The exact amount of electricity consumed by data centers, globally or in the United States, which hosts more than any other nation, isn't publicly reported by all tech companies, says Eric Masanet of the University of California, Santa Barbara, who researches data center sustainability. But according to the most recent estimates by the International Energy Agency, US data centers guzzled some 224 terawatt-hours of electricity in 2025 — more than 5 percent of the country's electricity use. That's a significant uptick from an estimated 1.9 percent consumed in 2018, well before the mainstream surge of generative AI.

This electricity use seems set to soar. In the race to secure market leadership for generative AI products, companies like Google, Meta, Amazon, OpenAI, Anthropic, Microsoft and Oracle are investing tens to hundreds of billions of dollars to build AI-focused data centers. Compared to data centers of the pre-AI days that consume, say, 100 megawatts of electricity — enough to power 83,000 homes with average demand — the newcomers are often "hyperscale" and can use a gigawatt or more, or roughly a tenth of the electrical capacity of Los Angeles.

Masanet and other experts have been alarmed to see much of this demand met by plants powered by fossil fuels, such as gas, whose burning releases planet-warming carbon dioxide. A key reason is that data centers are often constructed in places without abundant renewable energy sources like hydropower, geothermal, solar or wind.

Tech companies often offset emissions by investing in renewable energy elsewhere. But unless those clean energy plants make more energy than the data centers use, this strategy — at best — keeps CO₂ emissions of centers in stasis rather than reducing them to a net of nothing, important for halting global warming. "For every megawatt for which we install fossil fuel power," Masanet says, "it sets us back on our progress."

And that's not considering the resources spent on manufacturing the hardware that fills new data centers, or the impacts on communities living near them, which often suffer from air and noise pollution from gas plants and possible strain on local water resources, which are used to cool the data centers.

Although forecasts for AI's energy impact remain devilishly tricky, especially since the size of payoffs from investments in AI are uncertain, it's clear to experts that energy-saving strategies are urgently needed. Without them, according to one 2025 estimate, US data centers could soon be releasing the equivalent of 24 to 44 megatons of CO₂ annually, the latter equivalent to the annual emissions of Norway.

And so computer scientists and engineers are rethinking some of the power-hungry hardware and software that fuel AI. They're working to develop energy-saving algorithms and processor designs, and carefully considering where, and how, data centers are constructed.

"AI's energy cost is not an accident: This is basically a product of how our systems are built," says Fengqi You, an expert in energy systems at Cornell University. But with the right mix of solutions, he says, "we could really reshape the trajectory."

The roots of AI's energy problem

To comprehend AI's energy cost, it helps to understand large language models (LLMs) — the lifeblood of AI text generation tools such as chatbots and AI assistants — specifically, ones based on a design described in 2017 by the machine-learning laboratory Google Brain. This design, transformer architecture, can process text at lightning speed by simultaneously taking each word and weighing its relationship to every other word it sees. It "learns" which words go together by computing how strongly each word relates to all other words in a text, examining each word in many contexts. (A similar design is used for AI image and video generators.)

On a computational level, this happens by converting words or word fragments into numbers and performing additions and multiplications between them. Key to the speed is being able to do these calculations in parallel, made possible by graphic processor units (GPUs) — mostly manufactured by the company NVIDIA — originally invented for rapid 3D rendering of imagery during gaming.

The initial training of an LLM, required to learn all these relationships, consumes vast amounts of energy. Because each word it trains on must be weighed against all others in a given chunk of text, the number of computations the model performs — hence the energy required — increases quadratically relative to the length of text (i.e., doubling the length of text quadruples the number of computations). That adds up quickly given that most LLMs are trained on massive swaths of publicly available internet text. Some estimates suggest that training GPT-4 — the iteration of ChatGPT that launched in 2023 — guzzled between 50 and 60 gigawatt-hours of electricity, enough to power San Francisco for three to four days.

But experts are more worried about the energy costs of using the models to generate data once they've been trained, a process called inference. "You train once, then you inference for a billion people in the world," says Mosharaf Chowdhury, an AI systems expert at the University of Michigan who has been measuring the electricity usage of a handful of large language models that have been made publicly available.

This process is surprisingly inefficient: Each time transformer models generate a word — by selecting the one with the highest probability of following the previous word, given context — they put the query and partially written answer through the model. In doing so, they apply all of the parameters they've calculated during training to understand language patterns — which number in the hundreds of billions or even trillions.

"The fact that you have to do a lot of calculations for a single word to be added — that’s a problematic thing," says Günter Klambauer, an AI expert at Johannes Kepler University in Austria.

Tweaking AI software to save energy

This recognition has triggered interest in smaller language models specialized to specific tasks. These are trained more narrowly, have fewer parameters — say, tens or hundreds of millions — and perform substantially less computation than larger models. In one 2025 paper published by UNESCO, computer scientist Ivana Drobnjak of University College London and colleagues compared energy consumption of Meta's language model Llama-3.1 with smaller AI models dedicated to particular tasks — ones called DistilBART and t5-small-xsum for summarization, and others for translation or answering questions. When used for their respective tasks, the smaller models consumed more than 90 percent less energy than Llama 3.1 on the same job.

And so computer scientists have been driven to build a similar kind of task specialization into LLMs themselves. In "mixture of expert" models, only particular parts of one big model are activated for certain tasks. These parts "learn to handle different patterns in language," Drobnjak says.

This is thought to be one reason why R1, an LLM developed by the Chinese company DeepSeek, reportedly consumed significantly less energy than other models (independent experts have raised doubts about those figures). Udit Gupta, an expert in electrical and computer engineering at Cornell Tech, says that LLMs like Gemini or ChatGPT are similarly routing queries to more specialized sub-models. "There's a lot of work being done on how to assess the complexity of the query or task that's coming from users and then find the right model," Gupta says. (While Google spokesperson Ralf Bremer notes that the 0.24 watt-hours currently spent on processing median-length Gemini prompts is already 33 times more efficient than it was back in 2024, some experts suspect that processing queries with an LLM still consumes more energy than an equivalent web search.)

Scientists are also exploring different kinds of LLMs, to break what Klambauer calls the "quadratic curse" of transformer models.

One alternative, called a long short-term memory (LSTM) model, gets around this alarming energy increase by temporarily storing a kind of summary of the prompt that was inputted by the user plus the text generated so far, akin to recalling important plot points instead of an entire movie. That way, it only has to process the summary, rather than all the words in the full text to date, every time it generates a new word. This prevents LSTM's energy costs from skyrocketing as it responds to a query — using about 50 percent less energy than transformer-type models to process texts of around 8,000 words in length, Klambauer says.

LSTM models were developed in the 1990s but were abandoned because transformers could be trained much faster. But Klambauer says that recent advances have improved the performance of LSTM, now called xLSTM. He's working with the Austrian startup NXAI to further develop and optimize xLSTM, "because we think it's worth it for energy efficiency," he says.

But major tech companies have invested so many years and resources into developing transformer-based models that switching to other models would be costly, says Wolfgang Maaß, an AI and business informatics researcher at the German Research Center for Artificial Intelligence. "We have to see whether this becomes as dominant, or whether it finds a niche in the whole market."

Computing with wafers and light

Though experts say the fastest energy savings will come from software tweaks, some are also taking aim at the energy-hungry processing chips that fuel AI computations. Engineers have made chips increasingly efficient over time by packing more computing capacity into individual processors — reducing the energy required to shuttle data between chips that are working together to perform AI computations. Engineers have done this by shrinking the size of transistors — microscopic electrical switches that process data — inside the chips.

But because engineers are reaching the physical limits of how small transistors can be, "we need to think of alternate ideas to improve the designs," says computer architect Ajay Joshi of the Boston University Photonics Center.

One strategy is to make the chips larger. Dinner-plate-sized "wafer-scale chips" can pack nearly 70 times as many transistors as a single, postage-stamp-sized GPU and consume 143 times less electricity for communication than comparable GPUs, says computer engineer Rakesh Kumar of the University of Illinois Urbana-Champaign. Commercially produced by the California company Cerebras, wafer-scale chips have drawbacks, including a greater risk of damage during manufacturing. But because of their energy-saving and other beneficial features, "they would be very attractive to many hyperscalers and AI companies," Kumar says.

Many tech companies have improved energy efficiency by fashioning their own processors that are tailor-made for AI computations — such as Amazon Web Service's Trainium2 chip or Google's Ironwood Tensor Processing Units — according to statements from those companies. As for NVIDIA, the company's head of sustainability Josh Parker says its AI-specialized GPUs have come a long way from the ones used for gaming and are now designed to run AI tasks as efficiently as possible; other innovations, such as making the interconnections between GPUs more efficient, have also helped. "Over the past eight years, NVIDIA GPUs have improved 45,000 [times] in energy efficiency for large language model workloads," he says.

Engineers are also exploring alternative computing methods. Conventional AI processors calculate by encoding numbers in a binary system of ones and zeros, which is achieved by turning transistors on and off (representing the number 5, for instance, requires four transistors to represent the code 0101). But transistors can do more than function as binary switches allowing electron flow or not; they can also work as analog dials and hold intermediate voltages representing different numbers. That requires fewer transistors, and less energy, for computations. "People have known for decades that doing certain things in analog … can be a lot more energy efficient," Kumar says.

For example, electrical engineer Paul Manea of the German research institute Forschungszentrum Jülich and colleagues are working to develop devices called "gain cells" that are full of transistors working this way. Importantly, gain cells can both store the data required to process a query, and compute the answer. That overcomes another big energy bottleneck of conventional computing systems, where memory storage and computation occur on separate pieces of hardware.

That's especially problematic for transformer-based LLMs, because each time they generate a word, they must shuttle the query and partially written answer from memory to a processor. Manea and colleagues estimate that gain cells in lieu of traditional GPUs can reduce the energy guzzled by one of the most energy-consuming parts of transformer-based LLMs by four orders of magnitude. But it will take more refining before they can be more widely used, Manea says.

The notion of devices that both store and compute information is a key idea of "neuromorphic" computing, an up-and-coming field of computer engineering inspired by the human brain, which consumes orders of magnitude less energy than computers. Another brain-inspired invention is chips that encode information not in continuous data streams but — like human nerve cells — in the timing of voltage "spikes" propagating through the system. Allowing components to rest until they're needed "could potentially translate to less energy," says Eleni Vasilaki, an expert in bioinspired machine learning at the University of Sheffield in England.

Maaß, for example, is part of a team that received roughly $5.8 million from the German government to test neuromorphic chips, among other strategies, to reduce the energy required for AI models. Some brain-inspired chips are already commercially available, but the technology is still far from being attractive for mainstream computing, says nanoelectronics expert Tony Kenyon of University College London, whose team recently received $17 million from the UK government to develop neuromorphic computing.

Other scientists are developing chips that process information not with electrons but through the interaction of photons — particles of light — with matter (fiber-optic cables, which encode and transmit data as light pulses, are used around the world). With photons, more information can be transmitted at the same time, and signals can be altered much faster, says Elena Goi, a photonic computing researcher at Friedrich Schiller University Jena in Germany.

Several companies have developed chips that can perform some AI computations with optical methods, says Joshi; he recently estimated that manufacturing optical chips could consume up to an order of magnitude less energy than conventional ones of the same size. Joshi hopes that, "in 10 years, we would have a practical solution that can be deployed pervasively across the data centers."

Reshaping AI's energy trajectory

Even without reinventing how computers work, much can be done to reduce AI's impact not just on energy but also on water resources used for cooling data centers. Importantly, tech companies should reconsider where they build those centers, says energy systems expert You. Right now, existing US ones are concentrated in northern Virginia, which has limited water resources and renewable energy capacity compared with the Midwest, for instance. You recently estimated that better siting — along with energy-efficient hardware and software — could reduce future carbon and water footprints of US data centers by 73 percent and 86 percent, respectively.

Masanet adds that tech companies already with data centers across the country could at least train their models in strategic places. "Some companies like Google have been doing this: They shift their loads to follow renewables," he says. They also should address the electricity and resources spent on manufacturing processors for new data centers, as well as electronic waste as outdated tech is replaced every few years, he adds.

Minimizing e-waste by using hardware for longer periods and recovering old electronics is one of Amazon's sustainability strategies, according to a statement to Knowable Magazine; so is designing data centers in energy- and water-saving ways and investing in a slew of renewable and nuclear energy projects. "We'll continue to implement solutions that benefit our customers and the communities we operate in," says Brandon Oyer, Amazon Web Services' head of energy and water in the Americas.

Meanwhile, a press representative at Microsoft points to a number of sustainability initiatives the company has taken, including new cooling technologies, renewable energy investments and waste reduction. Google spokesperson Ralf Bremer emphasized the company's goal of reaching net-zero emissions across its operations by 2030 and replenishing 120 percent of the fresh water consumed by its offices and data centers by 2030. An OpenAI representative points to a press release outlining efforts to minimize water use and plans for solar energy generation at one of its campuses. Anthropic, Meta and Oracle did not respond to requests for comment by deadline.

Though tech companies are taking sustainability into consideration, their main objective is to rapidly build out data center capacity, says computer engineer Benjamin Lee of the University of Pennsylvania. He predicts that, eventually, they'll need to step up efforts to improve energy efficiency to reduce costs. Governments should help to accelerate this shift, Masanet says. So far, he and his team have counted nearly 220 policies introduced to address data center sustainability at the US state level, 18 at the federal level, and more from other countries, though not all were ultimately adopted.

"It's clear that governments around the world are beginning to take action," he says. However, he adds, "we also see some state and local governments with proposed policies that mostly aim to incentivize and accelerate data center builds."

AI's energy cost will ultimately be a balancing act: Will it save more resources through its problem-solving abilities deployed toward everything from finding cancer cures to improving logistics, than it demands? But though building a more frugal, energy-saving AI is important, so is carefully considering where AI is needed, Kenyon says. Is the world truly a better place, for example, with nonhuman "AI agents" providing customer support?

"I think it’s a common mistake, when a new technology comes in, to suddenly think, 'Well, everything has to adopt that new technology,'" he says. "That approach really isn't doing us any favors."

This article originally appeared in Knowable Magazine, a nonprofit publication dedicated to making scientific knowledge accessible to all. Sign up for Knowable Magazine's newsletter.

Plants, insects, and larger animals, like the forest’s white-faced capuchin monkeys, are well adapted to these changes. But in 2015, during an abnormally severe drought influenced by the El Niño–Southern Oscillation (ENSO), Perry, an evolutionary anthropologist at the University of California, Los Angeles, observed behaviors that once seemed impossible.

Under normal conditions "The [capuchin] mothers are quite devoted," she explained. "Now, I was seeing babies crying on the ground piteously. And the mothers just looking down like 'Too much trouble' and walking off, abandoning their infants."

"Even capuchins have their limits," Perry said. "And we need to start paying attention because all the weather predictions are saying that we're going to get more unpredictability and more climate extremes."

Monkeying around

Odd Jacobson, a behavioral ecologist at the Max Planck Institute of Animal Behavior, was a student at Lomas Barbudal in 2016, a year after this severe drought. His focus was on understanding how the study site's 12 different capuchin groups were moving through the forest. But now he's set out to investigate how else climate extremes may affect the behaviors and social structures of these monkeys.

In a paper published in Nature Ecology and Evolution, Jacobson and his coauthors — including Perry — analyzed how climate variability correlated to the 33 years of geolocation data they had on the capuchins.

Their first step was understanding how the size of each group was affecting the relationships between monkeys within the same group. To do this, they observed variables such as daily fruit intake, the size of the group's home range, and the distance the group traveled each day to find food.

Finally, to understand how monkey groups interacted, they used a "hierarchical social relations model," which allowed the scientists to predict how two different monkey groups would move through the forest and where their territories would overlap.

The team repeated this process, two monkey groups at a time, until they analyzed the interactions between all 12 monkey groups at Lomas Barbudal. Then, they added the climate-over-time layer to predict how the home range overlap and encounter rates (meaning moments where capuchins from two different groups engaged, often violently) would change with the seasons.

Strength (and weakness) in numbers

Generally, large monkey groups have advantages and disadvantages in the forest. One key advantage is the ability to control resource-rich areas, such as land with fruiting trees known as food patches. A key disadvantage is increased intragroup competition for food, meaning the daily fruit intake of individual monkeys was lower.

The researchers found that during climatic extremes, such as extremely wet or dry seasons, this intragroup competition intensifies, making the group less efficient at foraging overall. Behavior between groups changed with the climate as well. For example, in a typical dry season, large groups often overpower smaller ones to take over areas with more available fruit, such as along rivers.

But the new research found that this long-understood idea doesn't always hold true: During extreme climate events, like a dry season made even drier by the effects of El Niño, capuchins didn't try to hoard the higher-quality areas.

"We don't really know exactly why," Jacobson said. "Maybe there's not as much heterogeneity in the landscape during these resource poor times, and so there's not much that larger groups can monopolize."

Climate extremes, the research suggests, may be upsetting the balance that determines the optimal size of monkey groups. And, as a warming atmosphere makes climate extremes like El Niño or La Niña more intense, it's growing increasingly important to understand how these changes will affect animal societies.

Filippo Aureli, an ethologist at the Universidad Veracruzana, in Mexico, was not involved with this study, but he has studied the effects of extreme weather events on spider monkeys in Mexico. He also registered the infant mortality rates of capuchin and spider monkeys in the Costa Rican dry tropical forest during that 2015 drought. Capuchin populations experienced high infant mortality during the extreme event, while spider monkey populations tended to stop reproducing.

"With climate change, [climate extremes] are going to be more frequent and intense," Aureli said. "And we don't know what's going to happen. For this period [so far], they've held on very well, the spider monkeys, but we don't know for how much longer."

Perry agreed, noting "the importance of having a baseline when you're trying to study rare events like El Niño droughts."

"We know what normal is," she explained. "If you just try to drop in right now in all the chaos that we're starting to feel around the planet, then you really can't study it."

This article was originally published on Eos.org. Read the original article.

A photograph of Earth glowing in deep space, the moon's cratered horizon stretching across its foreground, caught many people's eyes in April 2026. Astronauts captured the image while aboard NASA's Artemis II mission, and like the famous Apollo 8 "Earthrise" image, the picture felt instantly real and inspiring for many.

But when almost anyone can fabricate a visually similar image in seconds from a text prompt using artificial intelligence, how do people decide which image is real?

The proliferation of AI-generated science images in public spaces is not simply a misinformation problem. As a researcher who studies visual science communication and public trust, I believe it also contributes to a crisis of trust in science in the age of AI, and the tools scientists have long relied on to establish visual credibility are losing their grip.

AI-generated images infiltrate science

AI tools are already changing how scientific visuals are created, shared and publicized.

Researchers use them to generate illustrations, create synthetic data, edit lab images and produce materials for education and public outreach.

While AI can help scientists communicate complicated ideas more creatively and efficiently, these same tools blur the lines between illustration, enhancement and fabrication.

In 2024, two papers were retracted after publishing AI-generated figures posessing biologically impossible structures. In April 2026, the New England Journal of Medicine retracted a paper after discovering that a clinical image had been manipulated with AI. These are just cases that came to mass public attention and are likely just the tip of the iceberg. Researchers have warned that AI-generated visuals pose growing threats in fields that depend heavily on visual evidence, such as materials science.

Academic publishers are beginning to adopt AI-detection tools. However, systems designed to detect fake images will almost always lag behind systems designed to create them. Many detectors can identify only image patterns they were trained to recognize. As new AI models emerge, developers must constantly obtain new data and retrain detectors to catch up.

The biggest concern are realistic-looking visuals that subtly distort scientific details while remaining believable enough to pass initial review.

Trust in scientific images

For decades, scientific images carried authority partly because they were difficult to produce. Creating microscope images, climate graphs and space photographs required expensive equipment, institutional resources and specialized expertise. Most people assumed such images represented true observations because very few people could make them.

Research in science communication, including my own, suggests that people judge scientific visuals using a few mental shortcuts. Does the image look technically sophisticated? Does it come from a trusted institution? Does it match what I already believe? Generative AI is undermining all three of these heuristics, or mental shortcuts.

Today, anyone can create a polished, scientific-looking image from a text prompt. Images are also detached from their original source when circulating online. When visual quality and institutional attribution become unreliable cues for judging the credibility of science images, people tend to fall back on something else: their own prior beliefs.

As a result, authentic scientific images that challenge someone's existing beliefs can now be dismissed as AI-generated, whereas fabricated images that confirm them are easily accepted as evidence. AI, in this way, may amplify motivated reasoning — that is, people's tendency to accept what they already agree with and question what they do not.

This shift matters because visuals have long served as evidence for scientific claims. Nonexpert audiences rely on images not only to see what scientists have discovered but also to develop an emotional connection and perceive credibility in the science being presented.

If audiences stop trusting visual evidence altogether, science loses one of its most powerful tools for public communication.

Transparency, not restriction

AI tools offer real benefits for researchers communicating their work to diverse audiences. The challenge is using these tools without quietly transferring AI's credibility deficit onto the science the images are meant to convey.

One practical path forward is for researchers to treat image provenance — where an image came from and how it was created — with the same seriousness they already apply to data provenance.

Scientists routinely disclose funding resources, study methodologies and conflicts of interest. Similar standards may now be necessary for scientific images. Was AI used to generate or modify this image? Is it a direct observation, a simulation or an illustration? What exactly does the image represent, and how was it verified? Can it be replicated by other researchers?

My colleagues and I found that people's familiarity with AI significantly shapes how they judge the credibility of AI-generated visuals. Those familiar with AI tools were more likely to view AI disclosure as a sign of transparency, and some rated clearly labeled AI-generated content as more credible than unlabeled content.

Transparency gives audiences the necessary context to evaluate what they are seeing, but it may not resolve every dispute about how images are made. Responsible use of AI-generated scientific images will require honesty, adherence to professional norms and the collective development of evidence-based standards across fields.

Why authentic images remain powerful

The original Apollo 8 "Earthrise" photograph of 1968 carries significant emotional impact. So do the Artemis II images of 2026.

What makes them meaningful is not simply their beauty. It is their traceable connection to scientific reality. When people look at these photographs of planets, they also know there are astronauts, physical cameras, documented missions and verifiable observations behind the images. In this sense, authenticity is a documented relationship between an image and the world.

In the age of generative AI, scientific institutions can no longer assume audiences will automatically trust their visuals. Trust now depends on transparency, documentation and clear communication about how visual evidence is produced.

Without guidelines and standards, science risks entering a world where every image can be questioned and no image carries inherent credibility.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Who is responsible for AI's output? Artificial intelligence (AI) companies like OpenAI maintain that they are not. In fact, their terms and conditions in 2023 stated that responsibility lies solely with the user. A German court disagrees.

On June 9, a Munich court (subject to appeal) ruled that Google can be liable for false claims produced by its AI summaries, drawing a sharp line between ordinary search results and machine-generated assertions. In other words, AI companies must be held legally responsible for the output that is created by their systems and pushed to users.

The court's logic was simple but profound: Search results point outward to sources, while AI summaries speak in Google's own voice. That distinction matters because it goes to the heart of what kind of speech deserves protection — and what kind is subject to legal scrutiny. The U.S. should follow the German court's lead. In the absence of such provisions, the entire burden of discerning truth from falsehood falls on the reader.

In the U.S., the First Amendment is intended to protect the right to speak, argue, persuade and offend. But freedom of speech is not free of caveats. It does not allow people to incite others to commit crimes, to threaten or to defame, for example. And if speech causes material harm, speakers can be held liable for those harms. When a company chooses to put a synthetic answer engine between users and the web, it is no longer merely hosting speech; it is producing an amalgamation of complex mathematical expressions that, outputted as text, resemble human speech. AI companies want this text to enjoy the same protections user-generated text has, while simultaneously dodging all the responsibility associated with being a speaker.

The roots of this dilemma go back to the 1990s, when the advent of online forums and social media created a new problem. Unlike traditional publishers, forum hosts needed to provide a platform for their users' voices, without being liable for what their users were saying. This problem was addressed with Section 230 of the Communications Decency Act, enacted in 1996. Section 230 was a bipartisan amendment written to preserve the internet as a space where ordinary people could speak (or post) without the forum host becoming liable for every third-party post.

That broad immunity reflected a democratic judgment: If the law made platforms responsible for all user content, many would censor aggressively or stop hosting speech altogether. This would limit free speech. Section 230 was meant to protect the ecosystem of human expression. In this sense, hosts of online spaces can be seen as providing a public square where speech occurs.

The lawmakers who passed Section 230 three decades ago could not have foreseen a world populated by chatbot-generated text. As such text increasingly leads to real-world harms, lawsuits are proliferating and tech companies are deploying a number of often-contradictory legal strategies to avoid culpability. In some cases, they are arguing that AI-generated text is not speech, but rather simply a tool, and that companies are therefore protected as "carriers," not "publishers" by Section 230's protection of a public forum for free expression.

But the companies deploy this argument only when it suits them.

In other cases, they are increasingly reaching for free-speech language to defend AI-generated text because free-speech protections provide broad legal immunity. For example, in a Florida wrongful-death lawsuit against Open AI (maker of ChatGPT), a plaintiff has alleged that the company’s chatbot pushed a 14-year-old to take his own life. OpenAI argued that the chatbot was protected by the First Amendment, though the judge dismissed that defense and allowed the case to proceed.

Neither of these arguments is convincing. AI companies are not merely providers of a public forum, as the words produced by their AI summaries and chatbots are generated by the company's products.

Similarly dubious is the claim that bots should be seen as equal participants in a public square. This is a category error. Free speech is a human right — it protects people as speakers and listeners in a democratic public sphere. Bots do not vote, deliberate, dissent, worship or participate in civic life. They generate text, but they do not possess a moral and political standing. Bots have no skin in the game.

What, then, justifies constitutional protection in the first place? Extending the strongest speech protections to machines would not defend liberty; it would confuse "botput" with free expression. It would, in actuality, extend the strongest free-speech protection to companies. But that requires a separate line of argumentation that ought to be agreed upon by society.

The Munich court's limited and nuanced way of governing "botput" provides a clear way forward.

Given its history with Nazism, Germany does not enshrine free speech quite the way the U.S. does. But the German court's arguments still provide a useful template for a future U.S. ruling.

The Munich court held that if a system simply points users to sources, it resembles traditional search and should continue to enjoy broad protection afforded to aggregators. If it synthesizes claims, imitates the tone of authority, and offers a single authoritative answer generated by an AI, it should carry corresponding duties of care that entail liability for the company.

The need for such safeguards is only growing. AI-generated summaries can be copied instantly, scaled globally, and repeated across interfaces until a falsehood becomes regarded as "truth." That is not a hypothetical concern; it is already happening.

Moreover, it is important to remember that the original intention of Section 230 was to insulate platforms from liability for third-party posts, not their own text.

This is not an anti-innovation argument. AI can be helpful, efficient and genuinely transformative. The law should encourage useful tools while insisting that the companies deploying them remain responsible for the foreseeable harms of their products.

We need clearer rules that keep the internet free for people while preventing machines from laundering falsehood into authority. The German ruling points toward that future. The sooner U.S. law and policy follow, the better chance we have of preserving our shared reality and a healthy democracy.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

As summer in the U.S. heats up, people become more diligent about protecting their skin from the sun. Another option for doing so will soon be available.

On June 9, 2026, the U.S. Food and Drug Administration approved the first new sunscreen ingredient to be permitted for over-the-counter consumer use in the U.S. since 1999 — a chemical called bemotrizinol.

Bemotrizinol isn't new — consumers in Europe and Asia have used it for decades. Some are hailing its long-overdue approval and arrival onto the U.S. sunscreen scene.

I am a biomedical engineer studying skin science — including the damaging effects of the sun's rays. To understand what bemotrizinol does and how it fits in with products already available to consumers in the U.S., let's take a tour of the physics of sunlight and sunscreens.

A short primer on sunlight

Our planet is irradiated by a yellow dwarf star 93,000,000 miles away that we fondly call the sun. It radiates light from its surface at a temperature of about 10,000 degrees Fahrenheit.

The Earth's atmosphere blocks most of the sun's radiation. Of the rays that get through, about half consist of infrared light — which gives you that warm feeling you feel on a sunny day — and 40% visible light, which you are probably familiar with as daylight.

About 10% of those rays are ultraviolet, or UV, light. UV light has the shortest wavelengths of the three types. That makes it the most dangerous — it's invisible and can damage living tissue.

Ultraviolet damage

Physicists further categorize solar UV light into several types, based on the wavelength, which is measured in nanometers. About 95% of it is UVA (315-400 nm) and 5% is UVB (280-315 nm). Sunscreens need to be able to block those rays from penetrating the skin.

The sun also emits two other types of UV light — UVC (200-280 nm) and vacuum UV (100-200 nm) — but these are stopped by the atmosphere, so sunscreens do not typically need to be able to block them.

Scientists used to think only UVB was harmful because UVB rays cause sunburns. But today, researchers know both types of UV can damage the skin.

UVB, with its shorter wavelength, has more energy, but UVA can penetrate the skin more deeply. And all UV can degrade the integrity of your skin, damage the structure of your DNA and cause skin cancer.

The only natural safeguard your body has against UV light is a microscopically thin layer of a pigment called melanin in your epidermis. The skin produces more melanin when exposed to the sun — that's what tanning is.

This extra melanin does protect the skin, but not fully. That's why protecting your skin with sunscreen is so important.

Sunscreens old and new

Sunscreens come in two different forms — mineral and chemical.

The first chemical sunscreen, developed in 1891, was an ointment made from quinine — a plant-derived compound that makes tonic water bitter.

Chemical sunscreens cover the skin in a transparent coating, acting like a solar sponge. They absorb UV photons and undergo a harmless chemical reaction, then dissipate the energy as heat. Bemotrizinol falls into this category.

Mineral sunscreens such as zinc or titanium oxide ward off the sun's rays by forming a protective film that also absorbs most UV light, but reflects some of it. Unlike chemical sunscreens, the film absorbs the light naturally, without a chemical reaction — which is why they are often visible as a white film on the skin.

Chemical sunscreens that have been available in the U.S until now combine ingredients like avobenzone, the most widely used UVA filter, with UVB filters such as octinoxate, octocrylene octisalate and homosalate. Working together, these substances protect the skin against the broad spectrum of ultraviolet rays.

These sunscreens are only effective for a short time because they are degraded by the chemical reactions they undergo, which means they must be frequently re-applied.

Another important element of sunscreen — whether mineral or chemical — is its Sun Protection Factor, or SPF. This number tells you how well a sunscreen prevents your skin from burning — in other words, what amount of UVB rays it absorbs.

An SPF of 2 would mean a sunscreen cuts your exposure to UVB rays in half, filtering out 50% of those rays. An SPF of 30 means the sunscreen lets just 1/30 of the rays penetrate your skin - which is 3.3%. So it blocks about 97% of the UVB rays.

Dermatologists generally recommend using a sunscreen with an SPF of at least 30.

Benefits of bemotrizonol

Bemotrizinol, while new to the U.S., isn't a new compound. European regulators approved it in 2000. Chances are, if you brought back sunscreen from a vacation in Mexico, Europe, Canada or South Korea, you may even have some laying around your house.

One benefit of bemotrizinol is its ability to filter both UVA and UVB rays, so it doesn’t have to be mixed with other products to do the job.

It has some other beneficial features as well. First, its molecules prefer to sit on the surface of the skin rather than being more readily absorbed into the bloodstream, which can occur for some formulations.

Such absorption has raised concerns that sunscreens might be harmful — though this has not been demonstrated in people, it may discourage some people from using it.

Bemotrizinol also does not degrade as readily in the sun than other chemical sunscreen products. That photostability means it can last for four to eight hours, rather than having to be applied every two hours or so.

Regardless of the type, as a skin scientist I can say with certainty that any sunscreen is better than none. Your skin does an excellent job protecting you from the world outside — so make sure you protect it in return.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Pets form an important part of many people's lives, providing meaningful companionship. However, our pets can sometimes also be a source of unwelcome pathogens and diseases, particularly if they frequently roam outdoors.

We are ecologists and a veterinarian who study wildlife health and the movement of pathogens among wildlife, domestic animals and people. If you let your cat outdoors, or if outdoor cats visit your yard, our recent findings may be relevant.

Zoonotic pathogens are organisms that can infect both animals and humans. From a pathogen's perspective, humans are just another animal host. Wildlife is often emphasized as a source of emerging disease for humans because there are vastly more wild animal species than domestic animal species.

However, even if a pathogen is capable of infecting people, it needs a way to reach us. Humans share more zoonotic pathogens with domestic animals than with wildlife, because domestic animals live close to us. Pathogens benefit even further if they can infect a companion animal.

In our newly published research, we compiled data from more than 400 studies to investigate how a cat's lifestyle, whether they're mostly indoors, outdoor-roaming or feral, affects that cat's likelihood of carrying pathogens that can infect people.

Across this compilation, there were nearly 100 pathogens detected in cats that are considered zoonotic and capable of infecting humans. Familiar examples are rabies, Toxoplasma gondii, roundworms and Salmonella.

Our research

We found that outdoor-roaming pet cats had three to five times the odds of carrying a zoonotic pathogen compared with indoor-only cats. More surprisingly, cats allowed to roam outdoors had similar odds of carrying at least one zoonotic pathogen as feral cats. Outdoor-owned cats carried fewer types of pathogens than feral cats, but the same pathogens that infect feral cats can also infect owned cats.

These risks become a large-scale problem because pet cats that roam freely interact closely with people, wildlife and other domestic animals. Across the studies we reviewed, about 60 per cent of owned cats had unsupervised outdoor access; in some regions, that rate exceeded 90 per cent.

Roaming cats hunt, interact with wildlife or other domestic animals, and move through environments contaminated with pathogens and toxins. Research suggests that cat owners may underestimate hunting by around 80 per cent, meaning that many prey captures and animal contacts go unnoticed.

These interactions are not uncommon and not limited to so-called pest species. Single-country estimates of wildlife killed by cats run into the billions, with more than 2,000 wildlife species documented as prey for domestic cats.

Cats hunt animals that can carry zoonotic pathogens, including rodents, birds and bats, many of which would otherwise have little direct contact with people. Owned cats might bring home rodents carrying viruses, and there are documented cases of cats bringing rabies-positive bats into homes. A cat returning home with prey can therefore create a pathway by which pathogens circulating in wildlife populations reach people.

In addition, it is not only owners who are at risk. Outdoor cats defecate in gardens, parks, playgrounds and other shared spaces, potentially leading to high contamination rates. One study estimated that outdoor cats deposited more than 60 tonnes [60 tons] of feces per 10,000 households each year.

Depending on the parasite, feces can contain hundreds to hundreds of thousands of parasite eggs that can persist in soil or water for months to years, which can infect people or other animals that come in contact with those eggs.

What cat owners can do

The most straightforward intervention is also the most economical and humane: prevent unsupervised roaming. That does not mean denying cats access to the outdoors. It can mean building "catios" or enclosures, leash walks, supervised time outside or other forms of contained outdoor access.

Veterinary care still matters. Treating existing parasitic infections and vaccinating against diseases like rabies are essential precautions, even for indoor cats. Since neither vaccines nor anti-parasitic treatment cover the full spectrum of wildlife-associated pathogens, managing exposure remains the more comprehensive protective approach.

The free-roaming debate is often framed as a false choice: either cats roam freely, or they are deprived of a natural life. That framing is misleading and inconsistent with how we manage other companion animals.

We do not assume dogs need unrestricted access to roads, neighbors' yards or to hunt wildlife to have good welfare. Indoor cats and cats with supervised outdoor access can live healthy, enriched and longer lives.

Policies and strategies that address how and where owned cats roam outdoors can help safeguard biodiversity, feline and wildlife welfare and public health. That is the central insight of One Health, that the same choices that protect ecosystems can also protect the animals and people who share them.

This article was co-authored by David Lapen, who works for and receives research funding from Agriculture and Agri-Food Canada.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

How much of a cat fan are you? Find out by taking our cat quiz!

Humans are the only species known to use fully symbolic language: a system capable of expressing abstract ideas, imaginary worlds and endless combinations of meaning. But how did we get there?

The origins of language have fascinated philosophers, scientists and storytellers for thousands of years. Despite all our advances in linguistics, archaeology and cognitive science, we still don't know exactly how language began.

That uncertainty hasn't stopped people from trying to solve the mystery. In fact, some of the earliest theories of language's origins are among the strangest and most entertaining ideas in the history of science.

Bow wow, ding-dong

In the 19th century, scholars proposed a flurry of curious theories to explain how speech first emerged. Many of these theories were given playful nicknames by the German philologist Max Müller, who intended them partly as satire. Yet the theories were genuine attempts to tackle one of humanity's biggest questions.

The most famous is probably the Bow-Wow Theory. This suggested language began through imitation of natural sounds. Early humans, according to this theory, copied the noises around them: animal cries, splashing water, thunderclaps and birdsong. Words such as "buzz," "hiss," "bang" and "splash" seem to support the idea because they sound like what they describe.

But there is a problem. Different languages hear the same sounds differently. English dogs go "woof" or "bow-wow," but in Turkish they go "hev-hev," while Indonesian dogs go "guk-guk." Even animal noises, it turns out, are filtered through culture and language.

And onomatopoeic words (words that imitate sounds) make up only a tiny fraction of our vocabularies. Most words sound nothing like their meanings. For instance, there is nothing inherently tree-like about the word "tree."

That brings us to the Ding-Dong Theory, which argued that sounds and meanings are naturally connected in some deeper, almost mystical way.

Some words do seem to fit their meanings uncannily well. "Mini," "teeny" and "itsy-bitsy" feel small and delicate. "Lump," "rump" and "plump" sound heavier and rounder.

Modern linguists call this sound symbolism. One famous experiment asked participants to match two nonsense words, "bouba" and "kiki," to two shapes: one rounded and one jagged. Most people matched "bouba" with the soft shape and "kiki" with the sharp one.

The effect is real, but it is limited. Most language still appears to be arbitrary, which means there is no natural reason why a particular sound should mean a particular thing.

Pooh-pooh, la-la, ye-he-ho

Other theories focused less on imitation and more on emotion and social interaction.

The Pooh-Pooh Theory proposed that speech began with instinctive emotional cries such as "ouch," "oh" or perhaps less publishable exclamations uttered after stubbing a toe. According to this idea, language evolved from spontaneous vocal reactions to pain, surprise, fear or joy.

Again, though, there are complications. Interjections vary widely across languages. English speakers say "ouch." Greeks say "aou." Czechs might exclaim "ach." Emotional sounds are not nearly as universal as they seem.

Then there is the wonderfully named Yo-He-Ho Theory, which suggested language emerged from rhythmic chants used during collective labor, like sailors chanting "yo-heave-ho" while hauling ropes, or workers singing together to coordinate physical effort.

The theory may sound quaint, but modern researchers do think rhythm, cooperation and social bonding played important roles in human evolution. Language is, after all, deeply social.

Another proposal, the La-La Theory, linked language to music. Charles Darwin entertained the possibility that speech evolved from musical calls used in courtship and emotional expression. Before humans spoke, perhaps we sang?

Some modern theories echo this idea. One hypothesis suggests that, as early humans began walking upright, parents increasingly needed to soothe babies from a distance. Sing-song vocalizations, cooing and proto "baby talk" may have helped strengthen emotional bonds and eventually paved the way for speech.

Gestures, symbols and brains

Today, most scientists think no single theory fully explains language origins. Instead, language probably emerged gradually through a combination of gestures, vocalizations, facial expressions, social cooperation and increasing cognitive complexity.

Some researchers argue that language began with gestures before shifting to speech. Others believe language evolved as a tool for social bonding, allowing larger groups of humans to cooperate and share information. Still others see language as tied to the evolution of symbolic thought itself: our ability to imagine, plan, remember and communicate abstract ideas.

Biology is also a factor. Humans have developed unusually precise control over the tongue, lips and vocal tract. We have evolved specialized brain regions linked to language processing.

But anatomy alone cannot explain language. Parrots can mimic speech sounds. Many animals communicate. None, however, appear to possess grammar and symbolism on the human scale. And, frustratingly, early language leaves no evidence behind. Spoken words don't fossilize.

That lack of evidence is one reason the topic became so controversial that, in 1866, the Société de Linguistique de Paris banned discussions about language origins altogether, dismissing the field as hopelessly speculative.

Of course, theories about language origins also appear in religion and mythology. In Greek mythology, the messenger god Hermes was associated with language and communication. In the Hindu tradition, the goddess of knowledge and speech Saraswati bestowed Sanskrit upon humanity. In the Judeo-Christian tradition, language was a gift from God, who enabled Adam to name the animals in the Garden of Eden.

These stories reflect something deeply human: our urge to explain where language came from, because language itself feels almost magical. Every theory of language origins captures a small piece of the puzzle. Imitation, emotion, rhythm, music, gesture, cooperation and symbolic thought probably all played some role.

But none can provide a complete answer. The truth is that language evolved so long ago, and likely so gradually, that we will never pinpoint a single moment when it began, unless someone invents a time machine.

The birth of language will probably remain one of humanity's greatest unsolved mysteries. Still, the theories themselves tell us something important. Humans are always trying to explain what makes us human. And language may be the most human thing of all.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Humans are not isolated, interchangeable biological machines. We seek medical care carrying complicated backstories, life experiences, and cultural perspectives that shape how we experience illness, communicate pain, and respond to treatment.

For decades, medicine ignored this fact and the cost fell heaviest on patients from marginalized groups. Study after study has documented the result: stark disparities in health outcomes that track closely with race, income, zip code, and immigration status.

Medicine has only recently begun to reckon with this blind spot. For the last few years, medical students across the country have been required to receive training in what are called "social determinants of health" to better understand how trauma, poverty, racism, and life experience shape what patients bring into the exam room and what they need from the people treating them.

But the board governing medical school accreditation no longer wants this training to be mandatory. This would be a big step back for medicine. Keeping these standards in place is essential for making a more effective medical workforce that can deliver the best care possible while doing no harm.

My experience with both sides of the medical system — as a second-year MD/PhD student and a patient — shows why this training is so necessary.

A few months after I turned 21, I went in for a routine checkup and was told by my physician that I needed my first Pap smear. I understood the importance of the test, but it wasn't something I had prepared for that day.

When the exam began, I flinched at the cold speculum. My physician noted my tension and said it was interfering with the exam. I tried to relax, but as a survivor of sexual violence, the sensation triggered memories I hadn't expected. I asked her to slow down and at some point, stop. Instead, one of the physician assistants held my legs open as the speculum moved further. As the exam wrapped up, the only sounds in the room were my occasional sniffles as I cried quietly.

My mother, too, has endured the experience of being unheard by physicians. An immigrant from Accra, Ghana, she had healthcare providers hear her accent and see her non-citizen status and make a quiet calculation: one that placed her judgment below theirs and created a lifetime of consequences she continues to deal with to this day.

At five years old, the phrases "structural violence" or "medical racism" were not yet a normal part of my vernacular, but I didn't need a college education to understand that my mother's immigrant identity shaped the care she received, or more accurately, the care that was withheld from her. Her experiences were the consequences of a system that delivers less attentive, less thorough, and less humane care to patients who are Black, low-income, or immigrant. It has done so for so long that the disparity reads as normal.

We've finally started to see a shift in that system, however slight. A year after the first Pap smear, I needed an additional one. Despite knowing this screening was essential for my health, I was still very hesitant and expressed my fears to my physician. This time, however, my physician used trauma-informed care throughout the procedure. She checked on me before we got started, narrated each step, and moved slowly so I could stop and breathe when I needed to. She gave me back the control and autonomy I had lost. Once we were done, she checked on me before letting me get dressed.

When the door closed, I let myself sob, not from pain, but because she cared. I mourned the version of me that wished she could have been granted that care the first time, and I found myself mourning my mother too, feeling a connection to her in the most unfortunate way.

The difference in my two experiences was not incidental. It is the result of training. Medical education now teaches future physicians to recognize how trauma, lived experience, and social context shape patient care, which collectively is called structural competency.

Such training equips future doctors to ask better questions, communicate more effectively, and avoid causing harm, even during routine procedures. Since starting medical school, I have seen how this training shapes the way my peers and I understand patients. A patient with uncontrolled asthma is not simply "noncompliant" with their treatment regimen, but may be living in substandard housing with mold exposure. A patient with poorly controlled diabetes may not lack "motivation," but may face food insecurity or unstable income.

Structural competency enables physicians to see the full picture and ultimately treat patients with the dignity they deserve. Rather than simply adjusting a medication dose or repeating discharge instructions, a structurally competent physician might coordinate with a social worker, connect a patient to community resources, or advocate for systemic changes that no prescription can address.

Research supports that structural competency training improves both how physicians practice and how patients fare. A 2016 pilot study in California found that such training led to better patient satisfaction scores, and a more recent study found that it increased residents' sense of competency.

The training curriculum has only been required for a few years, but some early evidence hints that it could yield positive patient outcomes. Physicians trained to recognize these factors are also more likely to advocate for policy changes related to housing, education, and income equality: key determinants that influence chronic disease prevalence, infant mortality rates, and overall life expectancy.

The American College of Physicians has formalized this thinking, publishing recommendations urging physicians to ask about and address social factors as a core part of patient care, and real-world implementation backs this up. One Michigan physician network that began systematically screening patients for social needs in 2017 conducted over 20,000 screenings and used the results to connect patients to social services.

This is partly why the Liaison Committee on Medical Education (LCME) that accredits med schools currently mandates that structural competency be taught in every school — but that's about to change. In the wake of the second Trump administration's attack on diversity, equity and inclusion efforts, the committee is indicating that it will likely no longer require structural competency training as a part of medical education.

It would be reassuring to think that even in absence of this training, nothing would change; that good physicians will remain attentive, and empathy will fill the gap. My mother's story and my own suggest otherwise. Medicine does not become neutral in the absence of this education. It defaults, reverting to entrenched habits of thinking that ignore the nuances of how a patient's background informs their care.

At a time when mistrust in medicine is already high, we cannot afford to strip away the training that helps rebuild it. So what can be done?

Medical students must advocate within their institutions to preserve and strengthen structural competency education. Faculty must keep such training in their medical curriculum, even if LCME doesn't require it. Most importantly, professional organizations from local institutions to national bodies like the American Medical Association, must speak out clearly and use their congressional lobbying powers to push for the LCME to reverse its decision.

I love medicine. I am training to become a physician-scientist because I want to help build a future where care is both innovative and just. But I will not stand by while medicine forgets its own history. We have made progress, inches in the right direction. So we must do all that we can to ensure that we don't fall back, because the consequences will ultimately be the trust, the support and the lives of our patients.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.

One argument often used to quell concerns about the rising energy and resource demand of data centers is that artificial intelligence (AI) models will need less in the future as they improve and become more efficient.

But this seemingly logical thinking is a trap, according to a new United Nations report that quantifies the environmental costs of AI.

The report estimates that by 2030, AI's energy use could double to consume 3% of the world's electricity, produce emissions to equal the UK and deplete more water for cooling than the annual drinking water need of the global population.

It also anticipates the use of AI will follow an economic principle known as the "Jevons paradox", which predicts that when technological improvements increase the efficiency of a resource, it leads to a rise, rather than a fall, in the total consumption of that resource.

The paradox is named after economist William Stanley Jevons who observed this effect with the use of coal in 19th-century England. Efficiency gains did not reduce overall consumption. Instead, the lower costs resulted in expanded use and higher overall demand.

As AI models become cheaper and more attractive, the report expects this to encourage new uses and higher volumes of use, eroding and possibly erasing any savings from efficiency advances.

To avoid falling into this trap, it lays out a roadmap for responsible AI use based on guiding principles of transparency, efficiency by design, equity and justice, lifecycle responsibility, global cooperation and sustainable use.

The scale of the problem

Last year, data centers already consumed as much electricity as Saudi Arabia, which ranks as the world's 11th largest electricity consumer.

If electricity use doubles as projected by 2030, the associated carbon footprint would require 6.7 billion trees grown over ten years to offset this demand.

Data centers would also require 9.3 trillion liters of water and land nearly ten times the size of Mexico City.

Beyond resource use, the report also underscores the structural inequity at the heart of the AI boom, with only 32 nations hosting AI-specific cloud infrastructure and 90% of that capacity located in the US and China.

It warns of a widening digital divide between nations that build and control AI systems and those that consume them, with the latter often bearing a disproportionate environmental burden caused by mineral extraction and e-waste.

Responsible AI use

Two main forces shape AI's operational footprint: how much we use it and how we use it.

This involves all tasks AI models perform, from text and code generation to image and video. Each of these tasks requires different levels of computational effort.

The model choice also matters as each AI system performs these task with distinct energy and environmental costs.

The report argues responsible AI requires full value-chain governance, from mineral sourcing to recycling and safe disposal.

It calls for a twinning of capability and environmental stewardship — thinking about both what AI can do for us and the protection of the natural environment.

This would mean making environmental disclosures a routine part of AI development, at both the model and task level, and incorporating projected AI demand in climate and energy planning.

Responsible AI is crucial as countries are promoting and adopting AI across government and the public sector.

In Aotearoa New Zealand, the government has launched a national AI strategy and a public service AI framework.

While the framework was informed by the OECD's values-based AI principles, including inclusive and sustainable development, there is no requirement for environmental disclosures and no regulator compiling energy use or emissions.

Likewise in Australia, improving public services is part of the national AI plan. For example, the National Film and Sound Archive of Australia has created Bowerbird, a machine learning-enabled mass audio and video transcription engine, to document material. The Department of Veteran's Affairs has developed a proof-of-concept tool to see whether AI can help speed up the processing of claims.

Both countries take a deliberate "light touch" and principles-based regulatory approach to AI. But this approach risks overlooking the growing environmental cost of AI that can't be solved by improving it.

The natural environment is foundational to the economy, culture and wellbeing. It should be at the center of our thinking. It’s time to rethink the AI innovation playbook and shift focus toward a sustainable tech future.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Around 1 in 5 Americans say they've seen a ghost. I'm not one of them, and I probably never will be. I blame my brain.

Let me explain. No one can say definitively that ghosts exist, but many people believe they do. Roughly three-quarters of Americans believe in some form of paranormal activity — not only ghosts, but psychic abilities, precognitive dreams, mediums and anything else that conventional explanations can't account for.

As a psychology professor, I often think about the subjectivity people use when interpreting experiences. I wonder, then, if there are perfectly ordinary explanations for seemingly extraordinary experiences. Maybe a perfect storm of everyday factors can converge and trigger the sensation of a paranormal experience.

In my new book, "Science of the Supernatural," I explore the idea that the human brain might be creating an experience of the supernatural by misinterpreting the external world. Here are three factors that might trick your brain into creating a fake ghost:

Haunted factor #1: Environmental stimuli

Anyone who's ever watched a ghost hunting show has seen the paranormal investigator mutter something like "The EMF's going crazy" when there's purported supernatural activity afoot. Electromagnetic fields, or EMFs, are invisible areas of energy created by electrically charged particles.

At present, there is no direct evidence that humans can consciously sense EMF the same way we can touch, see or hear things in our environment. But with a handheld device purchased at a local hardware store, you can measure them anywhere. An EMF detector picks up electrical or magnetic activity, whether human-made or otherworldly. But do EMF fluctuations relate to paranormal activity?

The scientific method might help answer this question. In one study, conducted in the South Street vaults underneath Edinburgh, Scotland, EMFs fluctuated more in areas with a history of ghostly happenings. Another study found greater variability of EMFs in the more "haunted" areas of Hampton Court Palace in England.

People might unknowingly be detecting changes in environmental stimuli, like electromagnetic fields. The question then becomes: Did the ghost cause the EMF, or did the EMF cause the ghost?

To date, only one research group has attempted to experimentally manipulate environmental factors, including complex EMF, and measure subsequent perceptions of the paranormal.

Participants did report many peculiarities, ranging from feeling dizzy to feeling like they were detached from their bodies and even sensing a presence — but these experiences didn't correspond to how the researchers varied environmental conditions, like EMF intensity. Interestingly, the people who described anomalous experiences were the same people who believed more strongly in the paranormal.

Do environmental factors like EMF lead to perceptions of the paranormal? On the one hand, there is a correlation between reportedly haunted places and EMF variability. And there are some indications that humans can detect magnetism. On the other hand, experimental manipulation of EMF did not relate to weird perceptions in a lab setting.

I think we need to look into other haunted factors.

Haunted factor #2: Neurological mix-ups

By applying a small electrical current to the side of the head, usually to evaluate a patient for a clinical procedure, researchers have observed some strange effects. One case study described a patient who experienced an "illusory shadow figure" that was mimicking, and even interfering, with their movements. Other people have reported out-of-body experiences.

Experimental evidence suggests that this brain area, the temporoparietal junction, is probably crucial for the feeling of embodiment — that you inhabit your own body. Disrupting this brain area seems to trigger a sensation of disembodiment.

Neuroscientists aren't completely sure how the sense of embodiment is built in the brain. The brain probably integrates bodily senses, like balance and position, with other internal processes, like a sense of self and agency. When this integration is altered, a person will experience very strange sensations.

Sometimes, misinterpretation of sensations from the body can happen during sleep, when your brain shuts out the external world. During rapid eye movement, or REM, sleep, when most vivid dreams occur, the brain sends messages that prevent movement of skeletal muscles. This inhibition causes complete paralysis during REM sleep. It is a neurological safeguard; without it, you would be likely to act out your dreams.

Some people, though, wake up during REM sleep and find that they cannot move. They may simultaneously experience rich hallucinations — the remnants of their dream. This experience passes quickly. But in that moment of sleep paralysis, the neural signals that control skeletal muscle movement are inhibited, resulting in a mismatch of feedback from the body to the brain. Most people respond to the missing sensory information with fear, which makes them more likely to experience the sights and sounds from their dreams as reality.

Haunted factor #3: Personality traits

Living through a paranormal encounter requires that a person label their experience as such. If a believer were exposed to fluctuating EMFs, for example, they might be quick to categorize the strange sensation as paranormal. A skeptic might note they felt weird or off, but probably not point to a paranormal explanation.

There's a growing body of research that suggests people with certain personality traits are more likely to believe in the paranormal.

For instance, some people are hyperaware of unconscious perceptions and ideas, which then permeate their consciousness. Often, these traits are associated with magical thinking, distorted or unusual thoughts, disorganized behavior and, sometimes, trouble forming close relationships.

Psychologists refer to this set of traits as schizotypy. They're related to schizophrenia, although being high in schizotypy doesn't mean you will be diagnosed with the disorder of schizophrenia. People with high levels of schizotypy are more likely to believe in the paranormal. They're also more likely to experience disembodiment and spontaneous sensory perceptions and have trouble discriminating between self and others.

All of these traits relate to the function of the temporoparietal junction — the brain area that helps you know you're located within your own body.

When haunted factors add up to a ghost

While I cannot say for sure whether ghosts exist, I can propose a plausible explanation for why some people might be more prone to apparent paranormal experiences than others.

Consider a person who believes in paranormal phenomena who experiences a natural change in electromagnetic fields or an episode of sleep paralysis. Those experiences induce unusual sensations that this person cannot explain. Searching for meaning in ambiguity, this person distorts their distinction between internally and externally generated sensations. They settle on the only explanation that makes sense to them — that this strange feeling they experienced was a ghost.

My guess is that belief in the paranormal is the glue that holds the haunted factors together to create the (mis)perception of a ghost.

One experiment asked participants to walk through a disused theater in Decatur, Illinois. Some were told that the theater was haunted, and some were not. Several participants noted weird sensations that they attributed to paranormal activity — but only those who believed that the theater was haunted reported these sensations.

Belief alone might not create a ghost, but belief combined with at least one haunted factor — environmental stimuli, neurological hiccups or psychological conditions — might be enough to make a ghost real.

This becomes a chicken-or-the-egg riddle — or in this case, the ghost or the EMF. Someone who is more likely to be sensitive to environmental factors or who experiences sleep paralysis might create belief from their experiences. When someone cannot explain these experiences with any "natural" explanation, a supernatural explanation might be the only one that makes sense.

I've never noticed EMF. I've never experienced sleep paralysis. I'm pretty sure I don't have personality traits like schizotypy. I don't believe in the paranormal. And I don't think I'll ever see a ghost.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Imagine receiving a test result that tells you your body is biologically five years older than your chronological age. You exercise regularly, get good sleep, eat healthy meals and have a happy personal life. What have you been doing wrong? Can this test be trusted?

Dozens of companies are marketing products that promise to reveal a person’s "true" biological age — that is, how well your body is functioning — for a price ranging from around US$30 to over $1,000. These products are based on epigenetic aging clocks, which are research tools that estimate age based on a person's DNA. These clocks are reshaping how scientists study aging and how the public thinks about it.

But while epigenetic clocks are highly effective research tools to study aging at the population level, they aren't designed to make claims about the health of individuals.

We are biobehavioral health scientists who study how early development and environmental factors across the lifespan shape biological aging, influencing health and disease decades later. As researchers who use epigenetic clocks in our work, we have found them to be highly informative tools when studying large numbers of people. But these clocks can provide faulty results at the individual level, and they do not meet the standards required of common medical tests.

What are epigenetic clocks?

Measuring reversible chemical changes to DNA, known as epigenetic marks, can provide information about how your body is aging.

Using DNA obtained from routine blood draws, researchers can measure millions of these epigenetic marks in an individual. Running statistical algorithms on this information can produce a single value that represents that person's epigenetic age, analogous to chronological age.

Epigenetic clocks work because the chemical marks on DNA can shift over time and are influenced by lifestyle, stress and the environment. These changes capture aspects of aging that chronological age alone may not reflect.

In this way, epigenetic clocks help scientists identify the experiences, exposures and behaviors that may accelerate or slow biological aging.

Not for individual health decisions

Why can't epigenetic clocks provide reliable results about biological age for individual people?

First, there are dozens of different types of epigenetic clocks, each designed for a specific purpose. Some are used to predict a person's age, while others are used to predict how fast someone is aging or how many years until they die. These different clocks do not always agree with one another, even when used on the same person.

Second, epigenetic changes are dynamic, making age predictions sensitive to short-term fluctuations in diet, environmental exposures, illness, time of day and other transient factors. As a result, estimated age could vary substantially depending on when someone is tested.

Third, constructing epigenetic clocks is technically challenging, and there is no established gold-standard method for generating clocks across laboratories. For example, testing epigenetic age in saliva versus blood samples can yield substantially different results for the same person. The technologies used to measure epigenetic marks have also evolved over time and will likely continue to improve. As these methods change, the original algorithms designed for specific measurement platforms may not perform the same way.

Fourth, scientists do not universally agree on what aging means, in part because it is a very complex process. Reducing that complexity to a single number, such as an epigenetic age, can be misleading.

Finally, epigenetic clocks are influenced by a person's history of trauma, discrimination and early life adversity. This makes their use at the individual level potentially problematic. On average, marginalized communities tend to show signs of accelerated aging when assessed with epigenetic clocks. If insurance companies began using epigenetic age estimates to set premiums, many people could face higher costs for biological differences shaped by circumstances beyond their control, potentially deepening existing health disparities.

Studying how aging unfolds over time

While epigenetic clocks are not appropriate tools for individual health decisions, this does not mean they lack value.

Researchers have used epigenetic clocks to discover lifestyle habits that can, on average, slow down aging. Some examples include reducing daily calorie intake, exercising regularly, maintaining a healthy diet, getting enough sleep and avoiding smoking.

Epigenetic clocks can also help test new drug therapies aimed at slowing down specific aging processes. For example, researchers have shown that rapamycin, a drug connected to various aging processes, can reduce the epigenetic age of human skin cells. There is also some evidence that a treatment designed to regenerate the thymus may slow or even reverse epigenetic aging after one year of treatment. However, researchers have seen these effects only when looking at groups rather than individuals.

Epigenetic clocks are helping scientists advance scientific research on the aging processes, but they aren't medical tests to measure individual health. In the future, epigenetic measurements may play a useful role in guiding personal health decisions. But for now, epigenetic clocks sold as biological age tests are best used and refined by researchers who are studying populations rather than individual people.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

If you have ever tried to repair something, realized that it was beyond your financial or technical means, and ended up buying a new one, you are not alone. Repairing electronics and household appliances has not been a real option in the United States for decades now, particularly for items that have proprietary software in them.

Absurd situations have proliferated. It can cost about the same to buy a new printer as it does to replace the ink cartridge. The U.S. Department of Defense cannot repair the weapons systems it purchases because the intellectual property rights remain with the manufacturer. John Deere, the farming equipment company, doesn't allow farmers to access the software needed to repair their own combines and tractors because, while the purchase covers the physical machinery, it does not cover the software.

One consequence, in addition to cost and frustration for consumers, is environmental harm. The U.S. is the world's second producer of electronic waste after China, to the tune of about 43 lbs (19.5 kg) of electronic waste annually per person. Only 25% of this e-waste is recycled.

The right-to-repair movement emerged in response, calling for people to be able to repair what they purchase, or have third parties do the repair work, without unnecessary financial, legal or technical barriers. Right to repair seems to be a rare area of bipartisanship in Congress. The Warrior Right to Repair Act — introduced in 2025 by a Democrat —and the Repair Act — introduced by a Republican — are two ongoing legislative initiatives to create a federal legal framework that would make it easy and cheap for American users to repair their devices. Both bills are fiercely opposed by industry groups.

As a scholar of American culture, I found through my research that the origins of the legal and technical obstacles to product repairs lie in debates in the 1980s over new media and copyright guardrails.

Hollywood and VCRs

The rapid rise and popularity of video cassette recorders, or VCRs, in the late 1970s transformed films and TV shows from transient experiences into tangible consumer goods. As I show in my book, "Videotape," despite the potential for extra revenue, Hollywood was alarmed by the fact that users were now able to copy films on videotape, and tried to stop the technology. Today's repair bans are part of that story.

The first U.S. copyright provisions were embedded in the 1790 Constitution. Over time, the law was amended to include new technologies, but at the core of future legal arrangements remained the initial intent: to protect the financial rights of creators while giving enough access to information for society as a whole to progress.

Until the second half of the 20th century, the American doctrine of fair use, which allows the unlicensed use of protected works under specific conditions, allowed judges to prevent copyright law from negatively affecting public interest. Organizations such as public libraries, book clubs, universities and news organizations benefited from this legal approach. The concept was codified into American law in the Copyright Act of 1976.

When the film studios took Sony to court to stop the production and sale of video recorders in 1976, they argued that Sony's product encouraged copyright infringement. But the U.S. Supreme Court ruled in 1984 that taping TV content for personal use did not violate copyright law, expanding the understanding of fair use.

The industry then focused on finding a technological solution to the piracy problem and on securing stricter legal protections for its products.

They identified the digital versatile disc, or DVD, as a safer alternative to the VHS tape. Initially, the DVD was a read-only format. It took a few more years of engineering before affordable recording was possible. Even then, the process was far more complicated for users than videotape recording. In 1997, barely one year after the video disc was launched, all of the Motion Picture Association of America member studios joined the DVD Forum, collectively adopted the new format and started to phase out films released on videotape.

Copyright and virtual locks

Then came digital rights management. Collectively, the term refers to the battery of technological tools that the industry developed in order to control user access to content. These include encryption software and various forms of authentication or enforcement software that limit which types of digital activities users can perform. For instance, some mechanisms block the option to download or share a digital file.

The Digital Millennium Copyright Act, or DMCA, signed into law by President Bill Clinton in 1998, provided the broad legal framework that allowed these technological locks to expand far beyond entertainment, including to software. The Digital Millennium Copyright Act reflected a new alignment in interests between the entertainment and software industries. It increased existing penalties for copyright infringement online and criminalized any technology used to bypass technological locks. The law was adopted although at the time — and since then — critics warned that it could stifle innovation and increase costs for consumers.

Since 1998, more and more consumer products, from toys to dishwashers, use microchips and proprietary software protected by copyright. Because of the Digital Millennium Copyright Act, third party repairers cannot alter or bypass the proprietary software. If they did so, they would be liable for infringing the manufacturer's intellectual property rights, as is the case for John Deere farm equipment. Some electronics are even designed to make tampering with the product impossible.

Manufacturers maintain that only they or authorized personnel can and should repair their products. These repairs are often quite costly. When getting a product repaired becomes almost as expensive as buying a new one, many consumers will choose to buy and throw repairable items away.

Rising resentment over repair bans

Technology tends to outpace existing legal arrangements. With over 80% of Americans supporting the right to repair, it remains to be seen when or if American law will catch up with the unexpected consequences of a law meant to protect the intellectual rights of the creative industries, but which is now hurting consumers' pocket books.

This edited article is republished from The Conversation under a Creative Commons license. Read the original article.

Generative artificial intelligence (AI) is erasing the line between reality and illusion to the point where seeing is no longer believing. We need a social and legal framework that will separate real-world images from those generated by AI, as well as technical innovations, such as universal "AI watermarks," that will help viewers immediately distinguish real images from fake ones. Without such a framework in place, we risk losing the trust that real-world photography brings. And that would be a disaster for democracy.

On June 6, 1944, Allied forces stormed the beaches of Normandy. The photographs that emerged — grainy, blurred, chaotic — did more than document history; they shaped it. For millions who would never see the battlefield, those images became the war — visceral proof of sacrifice, courage and collective purpose. They transcended language, collapsing distance between the observer and the event.

The same can be said of other defining moments. The lone figure standing before tanks in Tiananmen Square. The falling man from the World Trade Center. The lifeless body of 3-year-old Alan Kurdi on a Turkish shore. These images are not merely records; they are cultural touchstones. They form a shared visual substrate upon which public understanding — and, often, political will — is built. They allow societies to coordinate emotion, judgment and action at scale.

But what happens when that substrate erodes?

Advances in generative AI make it possible to create images that are not only realistic but emotionally compelling and contextually plausible. Unlike earlier forms of manipulation, which required skill and often left detectable traces, today's synthetic images can be produced rapidly, cheaply and at scale. They can depict events that never occurred and people who never existed, in scenes that nevertheless feel uncannily authentic. And AI image generators are getting better.

This shift introduces a profound epistemological problem. Historically, photographs have occupied a privileged position in our hierarchy of evidence. "Seeing is believing" is not just a cliché; it reflects a deep-seated cognitive shortcut that also transcends written and spoken language. While we have always known that images can be staged or edited, the default assumption is that photographs bear some causal connection to reality. Generative AI severs that link.

The risks are not abstract. In the context of war, synthetic images are being deployed as propaganda — fabricated atrocities attributed to an enemy, or staged victories designed to boost morale. For example, an image of an American radar system allegedly damaged by an Iranian drone strike that was widely circulated turned out to be fake., In domestic politics, they are being used to inflame racial tensions, fabricate protests, or depict public figures in situations that never occurred. For example, a fake image of a mug shot of Donald Trump has been widely disseminated.

The speed and scale of digital dissemination via social media means these images shape perceptions before the images can be verified or discounted. For example, a picture of 250 poodle mixes in captivity posted by an animal charity was dismissed as being fake. Yet, it was real.

This example also highlights a more insidious consequence that may emerge in a second-order effect: Once the public becomes aware that images can be convincingly faked, genuine images lose their evidentiary force. This is the "liar's dividend" — the ability of bad actors to dismiss authentic visual evidence as fabricated. In such a world, even the most compelling photograph can be met with skepticism, its truth value perpetually contested.

Democratic societies depend on a shared baseline of facts and experiences. While disagreement over interpretation is inevitable — and often healthy — there must be some common ground regarding what has actually occurred. Images have long played a crucial role in establishing that. When their credibility collapses, so does the capacity for collective judgment.

This is not a problem that can be solved through technology alone. While detection tools and forensic methods will continue to improve, they operate in an adversarial dynamic with generative systems. Each advance in detection is met with a corresponding advance in evasion. Moreover, technical solutions often struggle to scale across platforms and jurisdictions, and they require a level of public understanding that cannot be assumed.

What is needed is a societal and legal response that reestablishes trust in visual media. There is a historical precedent. In the 20th century, the rise of photography prompted legal innovations around authorship and ownership. Copyright law did not prevent manipulation or misuse, but it created a framework for attributing images to identifiable creators, thus enabling accountability and recourse where necessary. Broadly speaking, this framework makes it possible to sue for defamation, libel, etc.

A similar approach could be adapted for the age of generative AI. One element would involve mandatory disclosure: AI-generated images would be required to be clearly labeled as such, both at the point of creation and in downstream distribution. This could be enforced through platform policies and, where necessary, regulatory mandates. This would mean even an inattentive viewer would immediately know whether an image were AI generated.

More importantly, there is a need for traceability. Advances in cryptographic watermarking and content provenance systems offer a pathway. By embedding metadata that records the origin and transformation history of an image, it becomes possible to verify whether a visual artifact is authentic, synthetic or altered. Crucially, such systems would need to be standardized, interoperable and resistant to tampering.

Legal frameworks would need to support these technical measures. They could include liability regimes for the malicious use of synthetic media, as well as obligations for platforms to preserve and transmit provenance information. Just as importantly, there must be institutional actors, including journalists, courts and civil society organizations that are equipped to interpret and communicate this information to the public.

None of these measures will fully restore the epistemic status or "truth value" that photographs once held. The age of naive visual trust is over. But the goal is not to return to a bygone era; it is to construct new mechanisms of trust that are robust to the realities of digital manipulation.

The images of Normandy, Tiananmen Square and countless other moments continue to resonate because they are widely accepted as reflections of reality. Preserving that capacity — for images to anchor shared understanding — is not merely a technical challenge. It is a democratic imperative.

Opinion on Live Science gives you insight on the most important issues in science that affect you and the world around you today, written by experts and leading scientists in their field.