Did You Know news

Full text

Inside the world’s newest data centers, energy-guzzling computations proceed round the clock as AI chatbots and other generative AI tools tackle tasks from the frivolous to the weighty: assembling imagery for social media, proffering relationship advice, analyzing medical images to diagnose cancer, creating code for developers or detecting financial scams for banks.

Just as the popularity of AI tools has skyrocketed in recent years, so have the associated environmental costs. Data centers now consume 414 terawatt-hours per year, or about 1.5 percent of global electricity use, according to the International Energy Agency — an amount that grew by 12 percent annually for five years before jumping to 17 percent in 2025. By 2030, the agency projects that demand for electricity by data centers will more than double. Much of the increasing demand for electricity is being met by fossil fuels, while experts also worry about the use of local water resources to cool data centers in drought-struck regions.

Use of AI to generate text or imagery probably accounts for a mere sliver of any given person’s environmental footprint. And experts stress that the onus is on tech companies to reduce AI’s resource consumption , from creating smarter, energy-saving algorithms to building more efficient hardware.

Yet there are simple actions people can take to ensure that their AI usage has as little environmental impact as possible — from carefully considering where AI is needed to tailoring prompts to minimize the amount of computation required.

“Individual choices are not meaningless, and some are more powerful than people realize,” says computer scientist Ivana Drobnjak of University College London.

Knowable Magazine Knowable Magazine Knowable Magazine Knowable Magazine Knowable Magazine Energy-hungry bots

It is notoriously tricky to estimate the energy expended on processing an individual chatbot query. Google, for example, estimates that its chatbot Gemini consumes around 0.24 watt-hours to respond to a median-length text query — equivalent to the electricity needed to watch TV for less than nine seconds. It also uses about 0.26 milliliters of water and emits the equivalent of 0.03 grams of carbon dioxide (driving a gas-powered car for a mile would emit about 400 grams). Small individually, these expenditures build up for those individuals and companies that use AI tools a lot.

The reason AI models consume so much energy lies partly in the processors that power them, such as the graphic processing units (GPUs) that consume significantly more energy than the central processing units (CPUs) that fuel simpler tasks like web searches and email. It also has to do with the models that underlie most popular generative AI tools, including the large language models (LLMs) that power AI chatbots and assistants.

These are based on a particular design called transformer architecture. This allows LLMs to train on vast swaths of language patterns in text and, from this, compute hundreds of billions or trillions of parameters. These parameters can then be used to generate new strings of text, by predicting which words are likely to follow one other.

A transformer-based LLM is computationally intensive because for each new word it generates in response to a user’s query, it runs the query and the words that have been written so far through the model, performing billions of calculations each time.

Tech companies note that LLMs have become more energy-efficient over time; according to Google’s 2025 calculations, the 0.24 watt-hours that Gemini consumes on a median-length text prompt represents a 33-fold decrease compared with the model’s energy consumption the previous year.

In any case, even small amounts of energy add up quickly given the scale of AI use. Based on 2025 numbers from tech company OpenAI, Drobnjak estimated in May that, at that point, around 3.2 billion queries are being sent every day to its chatbot ChatGPT. Users are asking AI tools to process and produce vast quantities of text, images and video. Some are having lengthy conversations with chatbots. And, increasingly, people are creating their own “AI agents” that themselves send queries to AI chatbots.

So what can users do to minimize the resources spent on their AI use? Experts have some tips.

Don’t give up on search

As a first, simple measure to save energy, users should carefully consider whether they truly need AI for a given task. “Asking ChatGPT ‘What should I wear today?’ or ‘How is the weather?’ is like taking a Concorde to travel to your supermarket,” says Günter Klambauer, an AI expert at Johannes Kepler University in Austria.

The same goes for web search engines that use AI to automatically generate a response to a query alongside the actual search results, such as Google’s AI overviews or Bing’s Copilot search. “If you’re just looking for a particular article, turning that off could be powerful from a saving-energy perspective,” says Udit Gupta, an expert in electrical and computer engineering at Cornell Tech in New York City. Selecting “Web results only” in one’s browser or including “-ai” in the wording of your web search can do the trick.

Smaller models use less energy

People and businesses that use AI tools a lot for specific tasks like translating or summarizing could consider shifting to smaller language models that are specialized to these tasks. Because these are trained more narrowly and perform fewer computations, they expend less energy than the massive, all-purpose LLMs on the same tasks.

In one 2025 study published by UNESCO, Drobnjak tested the benefits of using small models — such as one called opus-mt-en-es for English-Spanish translations, and other models for summarization and query-answering — in lieu of the model Llama 3.1 developed by Meta. Though these smaller models are often less user-friendly than more popular AI models, they’re freely available from the AI platform Hugging Face. The small models consumed between 15 and 50 times less energy while producing higher-quality outputs on the tasks for which they were designed, the study found.

Using small, specialized models for particular tasks consumes a fraction of the energy guzzled by large, all-purpose models, with similar if even slightly better accuracy. (Image credit: Knowable Magazine) The shift away from larger models resulted in a 90 percent decrease in energy use overall, making this the most powerful single energy-saving strategy Drobnjak tested in her study. As Gupta puts it, “You don’t need to use a trillion-parameter model for editing an email.”

Less chatty chatbots

Because LLMs perform so many computations for every consecutive word they produce, it helps to choose models that produce less text in general. AI systems expert Mosharaf Chowdhury of the University of Michigan, who has been measuring the electricity usage of LLMs that have been made publicly available, has learned that models that are “chattier” by nature tend to consume more energy.

For instance, one version of the model Qwen developed by Chinese company Alibaba Cloud consumes significantly more energy when it’s in “problem solving with reasoning mode,” where it produced roughly 10 times as many words in response to a prompt compared to its “text conversation” mode. So some experts recommend using reasoning mode only for complex questions and otherwise sticking with a chatbot’s standard mode.

Simply asking AI chatbots to “be brief” or giving them a word limit can also save energy. In the UNESCO paper, Drobnjak and her colleagues found they could reduce the energy consumption of the Llama model by 50 percent when they instructed it to halve its output. By contrast, keeping the prompt itself short had less significant savings — just 5 percent for a prompt that was half the length of the original query.

“The size of the output is what determines and drives the energy expenditure the most,” Drobnjak says. She has collaborated with the city of San Francisco to develop energy-saving tips for AI users, which include being as specific as possible and adding instructions like “five bullets max.”

Keeping chatbot prompts short can conserve some energy, but asking chatbots to keep their responses brief amounts to much bigger savings. (Image credit: Knowable Magazine) Go low-res and batch video

Similar recommendations apply for generating images and video, which can consume orders of magnitude more energy than generating text, as they involve iterating millions of pixels many times over, each time processing the entire image anew, says Drobnjak. Such tools are highly popular: Nearly 40 percent of teens ages 13 to 17 surveyed in a recent study by the Pew Research Center use AI to create or edit images or videos.

Drobnjak recommends generating images or videos only when necessary and only at the resolution necessary. “One option is to just start in low resolution,” she says, “and if the algorithm is in the right direction, you then start increasing resolution.” She also notes that editing existing images is always less computationally intensive than generating new ones from scratch.

And when generating multiple images, it helps to do so in a single session or batch, which is more efficient than doing so in multiple separate requests.

These actions may seem like a drop in a bucket, and in many respects they are, experts say. But small things add up. While waiting for tech companies, scientists and policymakers to find ways of reducing AI’s overall environmental impact, “the individual who knows to reach for the right tool can make a real difference,” Drobnjak says. “AI is [consuming] so much energy that we have to look at it from every angle.”

Editor’s note: This story was updated on July 21, 2026, to clarify that the energy use of individuals who use AI tools a lot is cumulative, not necessarily huge, as was originally stated.

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 .

Full text

Millions of people across Europe and North America will witness a solar eclipse on Wednesday, Aug. 12, but the event will vary greatly depending on the location. A narrow corridor stretching from Greenland to northeastern Spain will experience a total solar eclipse , much of Europe will enjoy a deep partial eclipse, and parts of North America will see a smaller partial eclipse.

Want to know what you can expect to see from where you are? Here's an overview of the event from various locations, along with the exact viewing times for the Aug. 12 solar eclipse.

Remember, it's important to wear certified solar eclipse glasses during all partial phases of the eclipse. Totality ‪—‬ when the sun's face is completely blocked by the moon ‪—‬ is the only time it's safe to remove your eclipse glasses. (It’s not too late to order last-minute eclipse glasses if you don’t have a pair on hand.)

If you're in the path of totality

A map of the path of totality (red) for the Aug. 12 solar eclipse. (Image credit: NASA's Scientific Visualization Studio) If you're fortunate enough to be inside the 180-mile-wide (290 kilometers) path of totality — the path of the dark shadow of the moon — in eastern Greenland, western Iceland or northeastern Spain, you'll experience the moon completely covering the sun for up to 2 minutes, 18 seconds. The sky will darken to twilight, bright stars and planets may appear in the daytime sky, and the sun's spectacular corona (its wispy outer atmosphere) will become visible. Here’s a handy guide to all the gear you’ll need to get the most out of totality.

The closer you are to the centerline of the path, the longer totality will last. But the most important consideration is to stay within the boundaries of the path, because it has a hard edge. Madrid, for example, is barely outside the path and will see a 99.99% partial eclipse — and no totality whatsoever. The partial phases last an hour on each side of totality.

Here's when totality will occur in various locations for the Aug. 12 total solar eclipse:

Scoresby Sund, Greenland: 4:35 p.m. WGST (up to 2 minutes, 17 seconds)

Reykjavik, Iceland: 5:48 p.m. GMT (60 seconds)

León, Spain: 8:28 p.m. CEST (1 minute, 45 seconds)

Soria, Spain: 8:29 p.m. CEST (1 minute, 43 seconds)

S'Arenal, Mallorca, Spain: 8:31 p.m. CEST (1 minute, 36 seconds)

If you're in North America

Viewers in North America will see a modest partial eclipse, with the biggest eclipse seen from northern and eastern Canada, Alaska and the northeastern U.S. The farther northeast you are, the deeper the eclipse will appear. Eclipse glasses are essential at every stage of the two-hour event, because there will be no totality. Here are the times of maximum eclipse in various locations:

Iqaluit, Nunavut: 1:24 p.m. EDT (61% eclipse)

St. John's, Newfoundland and Labrador: 3:00 p.m. NDT (53%)

Fairbanks, Alaska: 8:27 a.m. AKDT (37%)

Anchorage, Alaska: 8:21 a.m. AKDT (28%)

Bangor, Maine: 1:53 p.m. EDT (23%)

Quebec City: 1:46 p.m. EDT (24%)

Montreal: 1:45 p.m. EDT (18%)

Boston: 1:55 p.m. EDT (16%)

New York: 1:54 p.m. EDT (9%)

If you want to watch totality from outside the path, there are many live stream options available, including planned streams from NASA and the European Space Agency .

A deep partial eclipse will be visible in many locations throughout Europe, with a smaller eclipse visible in North America. (Image credit: GAGAN NAYAR via Getty Images) If you're elsewhere in Europe and North Africa

Most of Europe and far northwestern Africa will see a spectacular deep partial eclipse, with more than 90% of the sun covered in many western cities. Although the eclipse will never become total, it will still be an impressive sight, though solar eclipse glasses will be required throughout the event. Here are the times of maximum eclipse in some major cities:

Madrid: 8:32 p.m. CEST (99.9%)

Barcelona: 8:29 p.m. CEST (99.8%)

Algiers, Algeria: 7:33 p.m. CET (96%)

Nice, France: 8:24 p.m. CEST (95%)

Lisbon, Portugal: 7:36 p.m. WEST (94%)

Dublin, Ireland: 7:10 p.m. BST (94%)

Cardiff, Wales: 7:13 p.m. BST (93%)

Belfast, Northern Ireland: 7:08 p.m. BST (93%)

Paris: 8:17 p.m. CEST (92%)

Tangier, Morocco: 7:40 p.m. WEST (92%)

London: 7:13 p.m. BST (91%)

Edinburgh, Scotland: 7:05 p.m. BST (91% eclipse)

To check exactly what you'll see and when, enter your location on an interactive map of the eclipse , which will give you a schedule for when the eclipse begins, peaks and ends where you are.

Full text

Bolstering a type of bacteria that lives in the uterus may help counter age-related changes that can make it harder to get pregnant, a new study suggests.

While fertility decline is often linked to changes in egg number or quality , the uterus also becomes less receptive to an implanting embryo with age. The new research suggests that changes in the bacteria living there may be part of the reason.

In experiments, treating aged mice and lab-grown human uterine cells with a specific species of Lactobacillus , a group of bacteria often found in the uterus, reduced signs of aging in the uterine lining and improved embryo implantation. This improvement was also seen with a specific byproduct of Lactobacillus .

The findings, published Thursday (Aug. 6) in the journal Cell Host & Microbe , suggest that the uterine microbiome could be a promising target for treating age-related infertility. However, experts who were not involved in the study cautioned that the research is still preliminary.

"This is an interesting and promising approach, but we still don't know whether these findings will translate into meaningful improvements in human fertility," said Dr. Alex Robles , a fertility specialist at Columbia University Fertility Center who was not involved in the study.

"The microbiome can be influenced by many factors, including genetics, diet, geography, environmental exposures and other lifestyle practices," he told Live Science in an email, "which is why it's important to validate these findings across diverse populations before developing broadly applicable treatments."

To explore whether and how the uterine microbiome changes with age, the study researchers analyzed samples of the endometrium — tissue from the lining of the uterus — from 149 women ages 20 to 45 who were undergoing fertility treatment in China. The older women had lower levels of beneficial Lactobacillus bacteria, particularly L. gasseri and L. crispatus . These bacteria are generally associated with a healthy reproductive tract .

With age, the lining of the uterus typically becomes less receptive to implantation, lowering the chance of pregnancy. (Image credit: BSIP / Contributor vis Getty Images) "One finding that surprised us was that overall microbial diversity did not change markedly across the different age groups," study co-author Dr. Rong Li , deputy director of the Department of Obstetrics and Gynecology at Peking University Third Hospital in Beijing, said in a statement . "Instead, the major differences involved particular bacterial species."

Later, 93 of the women underwent in vitro fertilization (IVF), and of those, 56 became pregnant. Those who became pregnant had higher levels of Lactobacillus in their uterine lining than those who did not get pregnant. That said, the pregnancy rates were similar across women of different ages who underwent IVF.

To better understand the effect of these bacteria on the uterus, the team isolated five Lactobacillus species and tested how they interacted with uterine cells. L. gasseri stood out because it attached strongly to uterine cells, and this attachment was associated with reduced inflammation and lower damage from oxidative stress . The bacterium also produced large amounts of exopolysaccharides (EPS), a type of sugar molecule.

The researchers then purified these sugars into a postbiotic they call LG-EPS. Unlike probiotics , postbiotics are products of bacterial cells and thus do not contain live bacteria.

Then, the researchers ran tests in older female mice (roughly equivalent to a middle-aged woman) to see whether introducing L. gasseri or LG-EPS to the vaginal canal could affect the aging uterus. After four weeks of daily vaginal treatment with either the bacterium or the postbiotic, the mice had fewer signs of uterine aging, less inflammation and higher embryo implantation rates than the untreated animals did. (That said, the team did not measure how much of each treatment actually reached the uterus.)

When they tried the same test with a modified strain of L. gasseri that could not make EPS, they found that many of these benefits disappeared. That suggested that the exopolysaccharides were responsible for much of the benefit. LG-EPS also reduced signs of aging in artificially aged human uterine cells, improving their ability to support embryo implantation.

Further experiments suggested that the postbiotic works by modifying a signaling pathway that helps control cell growth and tissue repair. This pathway was overactive in aged mice and uterine cells, but treatment helped to normalize its activity. Blocking a key protein in the same pathway eliminated the treatment's effects.

RELATED STORIES

The choice of sperm is 'entirely up to the egg' — so why does the myth of 'racing sperm' persist?

1st 'atlas' of human ovaries could lead to fertility breakthrough, scientists say

'Useless' female organ discovered over a century ago may actually support ovaries, study finds

Robles said postbiotic treatments could be especially interesting because they may be easier to manufacture than treatments containing live bacteria. "If confirmed in future studies, these treatments could potentially be used alongside IVF to improve the uterine environment before embryo transfer," he suggested.

But he cautioned that the study should not prompt people to try over-the-counter probiotics or other supplements as ad hoc fertility treatments.

"The reproductive tract is a delicate ecosystem, and introducing or eliminating certain bacteria could have unintended consequences," Robles said. "We do not fully understand what constitutes a 'healthy' uterine microbiome. As such, any intervention should be approached with caution."

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

Full text

Your favorite stretchy workout clothes may be much easier to recycle one day, thanks to a new type of elastic yarn that is as strong and stretchy as today's spandex-containing fabrics.

The yarn, described in a July 25 study in the journal ACS Materials Letters , is built entirely from polyethylene , the plastic used in milk jugs and single-use grocery bags. By contrast, most of today's stretchy fabrics blend spandex, a bouncy-but-weak polyurethane fiber, with a tougher fiber like polyester or nylon.

That combination gives clothes their stretch and durability. But because the fibers are chemically different, it's difficult and costly to separate them for recycling , and chemicals used to break down the fabrics are toxic. The result is that almost none of this material is recycled. An estimated 80% of textiles sold in the U.S. contain spandex, and most end up incinerated or in a landfill, study co-author Svetlana Boriskina , a research scientist in the Department of Mechanical Engineering at MIT, said in a statement .

To solve that problem, the MIT team turned to polyethylene, one of the world's most widely recycled plastics . For the yarn's stretchy core, the researchers created a soft, springy version of polyethylene akin to a rubber band. Around that core, they wound a second, stiffer polyethylene fiber, more like a fishing line, coiling it tightly the way a spring or a twisted phone cord is coiled. Because both materials come from the same plastic family, they don't need to be separated before they're recycled.

RELATED STORIES

Scientists break down cheap plastic using the air — and turn it into something far more valuable

Humans inhale a staggering amount of microplastic every week. Here's where it ends up.

How much plastic actually gets recycled?

In laboratory tests, the researchers melted a spool of the yarn and respun it into new yarn 10 separate times. Each time, they tested its strength, and the recycled yarn performed just as well as the original in every case. In head-to-head tests, the new yarn even outperformed commercial spandex-polyester yarn in maximum load, meaning it could bear more force before breaking, according to the study.

The team has already shown that these lightweight polyethylene fibers can be engineered for other perks, too ‪—‬ like cooling, moisture wicking and stain resistance ‪—‬ building on their earlier work spinning polyethylene into wearable fabric. Even buttons, zippers and buckles could eventually be made from compatible plastics, so a whole garment could be melted down and remade without being taken apart first.

Full text

Scientists are preparing for an interstellar rescue of humanity's most distant spacecraft, Voyager 1, after fixing up its twin from billions of miles away.

According to a NASA statement released Aug. 4, a clever power-saving maneuver known as the "Big Bang" helped the agency avoid having to shut off another one of the Voyager 2 probe's three remaining science instruments later this year. The Big Bang procedure involves selectively shutting off some high-power instruments on the Voyager probes while switching on lower-power alternatives, all while keeping the spacecraft warm enough to operate in the frigid environment of interstellar space.

"The idea is to swap out a group of powered devices all at once — hence the nickname — turning some things off, and replacing them with lower-power alternatives, to keep the spacecraft warm enough to continue gathering science data," representatives from NASA's Jet Propulsion Laboratory, which manages the mission, wrote in an April blog post .

NASA carefully ran tests on Voyager 2 before implementing the procedure in May and June. Now, sharing the results of those tests in the Aug. 4 statement, NASA reports a successful power transfer that should keep Voyager 2 running a while longer.

"The savings should provide power to keep its three instruments operating for at least an extra year," NASA wrote of the Big Bang.

But power is dwindling on both nuclear-powered spacecraft, and Voyager 1 is in worse shape than its twin. Mission scientists were forced to turn off one of Voyager 1's last remaining science instruments in April.

Voyager 1's turn for a life-extending Big Bang will happen within a few months, to keep science operating there as long as possible, agency officials added.

Going out with a bang

Scientists decided to do the "Big Bang" on Voyager 2 first because it is in better health, and closer to Earth, than Voyager 1. Voyager 1 is nearly 171 astronomical units (AU, where 1 AU is the average distance between Earth and the sun) from our planet, while Voyager 2 is about 143 AU away. Both spacecraft are crucial for revealing clues about the region outside the heliopause ‪—‬ the sphere of influence the sun exerts upon our solar system ‪—‬ as they are the only working spacecraft at such a distance.

That distance is no small factor. In November, Voyager 1 will reach a milestone "light day" from Earth — when it takes 24 hours for a radio signal to travel one way from our planet to the spacecraft. Both spacecraft will also turn 50 years old in 2027, assuming they keep working.

Both Voyager probes launched in 1977, with Voyager 2 getting an opportunity for a "grand tour" of our solar system. An alignment of the outer planets — Jupiter, Saturn, Uranus and Neptune — that occurs only once every 176 years allowed both Voyager probes to maneuver from one planet to the next with minimal fuel usage.

Voyager 2 reaches interstellar space. Here's what the spacecraft finds.

NASA switches off Voyager instruments to extend life of the two interstellar spacecraft: 'Every day could be our last.'

NASA engineers finally fix Voyager 1 spacecraft — from 15 billion miles away

Voyager 2 flew by all four planets, finishing its Neptune encounter in 1989, while Voyager 1 was tasked with flying above the plane, or "ecliptic," of our solar system after its closest approach to Saturn in 1981. As each spacecraft finished its close-up imaging of these planets, some instruments were shut off to extend their mission for as long as possible. But as the two spacecraft aged, more instruments needed to go quiet.

Plutonium, which the spacecraft rely on for power, has a natural decay rate, with the spacecraft now losing roughly 4 watts of power a year, according to NASA. In recent years, power supplies have turned critically low on both spacecraft. Voyager 2 would have needed to shut off one of its remaining three science instruments later this year if measures were not taken, and Voyager 1 was in slightly worse shape after an unexpected power drop in February occurred following a planned maneuver in deep space. Only two of Voyager 1's 10 science instruments remain powered after the latest was shut down in April.

Full text

Artificial intelligence (AI) has helped uncover new weaknesses in two cryptographic systems, including a simplified version of the Advanced Encryption Standard (AES) that underpins much of today's internet. While these headline-grabbing findings don't put anyone's passwords or bank accounts at immediate risk, experts say the research could mark the beginning of a new era in which AI becomes a powerful assistant for discovering flaws in the mathematical foundations of digital security.

In a blog post published July 28, representatives from Anthropic's Frontier Red Team said Claude Mythos Preview independently developed new cryptanalytic techniques against two different targets: a version of AES-128, one of the world's most widely used encryption algorithms, and HAWK, an experimental post-quantum digital signature scheme currently being evaluated as part of the U.S. National Institute of Standards and Technology's (NIST) effort to standardize cryptography for the quantum computing age .

It's tempting to interpret this news as AI cracking one of the internet's most important encryption algorithms, but the reality is less dramatic. Nonetheless, experts say the achievement could mark the start of a new age of digital security.

The power of encryption

AES protects huge amounts of everyday digital life, from encrypted websites and messaging apps to Wi-Fi networks and financial transactions. But the version Anthropic attacked wasn't the full AES-128 algorithm used in those systems. Instead, the researchers studied a seven-round version of AES, a deliberately weakened variant long used by cryptographers to test new attack techniques.

The full AES-128 algorithm uses 10 rounds of encryption, while Anthropic's research focused on a seven-round version. In a self-published study that has not been peer-reviewed, scientists Milad Nasr and Nicholas Carlini said Claude found a faster way to recover the encryption key from that simplified version, making the best-known attack between 200 and 800 times faster. However, the study stressed that the technique does not work against the full version of AES used to protect real-world systems.

The more significant result may instead involve HAWK. Unlike AES, which has protected data for more than two decades, HAWK is a relatively new digital signature scheme designed to resist attacks from future quantum computers . It's one of the remaining candidates in NIST's additional post-quantum signature standardization process, meaning it is still being scrutinized by researchers before its widespread deployment.

This development illustrates how AI can act as a powerful accelerator in cryptanalysis.

Thomas Espitau, head of research at PQShield

In a separate study , Anthropic scientists Zygimantas Straznickas and Stephen A. Weis found that Claude spotted a mathematical property that researchers hadn't previously exploited. Combined with existing attack techniques, this property made it much easier to recover HAWK's secret key.

Thomas Espitau ‪—‬ head of research at PQShield, a cybersecurity company that specializes in post-quantum cryptography, and a cryptographer who has published research on HAWK ‪—‬ described the work as "one of the most, if not the most significant, cryptanalytic result of the year."

"To say it plainly, this is great work, and it is exactly what the NIST process is designed to produce," Espitau told Live Science. "Candidate schemes exist to be attacked before they are deployed, not after."

Espitau said the attack combined previously known techniques with one missing mathematical insight that Claude identified, substantially reducing HAWK's estimated security margin. He believes the work demonstrates how AI could increasingly help researchers evaluate the strength of cryptographic systems before they are adopted.

"This development illustrates how AI can act as a powerful accelerator in cryptanalysis," he said. "Claude Mythos Preview identified the exploitation of the sign-flip symmetry, providing the final piece of a puzzle that the research community had been assembling."

Building defensive measures

However, not everyone thinks the announcement means AI suddenly outsmarted human cryptographers.

"There is nothing you need to change," Roberta Faux , head of cryptography at cybersecurity and quantum encryption company Arqit, told Live Science. "Both Anthropic and the outside cryptographers agree that you don't need to change your key sizes or switch your cryptography."

Instead, Faux said the bigger story is AI's ability to apply expert-level cryptanalysis across thousands of algorithms that relatively few researchers have had time to examine.

"The interesting prospect is not a model outdueling the world's best lattice theorist on one problem but a model applying solid, roughly expert-level analysis at scale to the several thousand ciphers nobody ever had the human-hours to examine," she said.

Related stories

Science word of the day: Cryptology

Quantum computing will make cryptography obsolete. But computer scientists are working to make them unhackable.

Scientists create new type of encryption that protects video files against quantum computing attacks

Faux also cautioned against attributing the HAWK breakthrough solely to AI.

"The HAWK attack used no exotic ingredients; it simply competently assembled tools that were already lying around," she said. "A post-quantum candidate is seriously scrutinized by maybe a few dozen people on the planet, so beating two years of review mostly reveals how thin that layer of review is."

For consumers, the immediate implications are reassuring. The encryption protecting online banking, shopping, messaging and cloud storage has not suddenly become obsolete. But for the researchers designing the next generation of cryptography, Anthropic's work hints that AI could soon help cryptographers test tomorrow's encryption schemes more quickly and more thoroughly than has previously been possible.

Full text

New images captured by the world's most powerful solar telescope have revealed something that was hiding in plain sight on the sun for 150 years: thousands of tiny whirlpools twisting across our star's surface. The ultra-detailed observations could help to explain some of the sun's most violent behavior as well as hint at the answer to a long-standing solar mystery.

The new time-lapse footage offers the first direct proof of a long-predicted phenomenon called Kelvin-Helmholtz instability . The effect itself isn't new to science; it's named for physicists Lord Kelvin and Hermann von Helmholtz, who described it in the 19th century . The physicists found that whenever two fluids slide past each other at different speeds, the friction between them, or shear, rolls into spiraling patterns. However, this is the first time this effect has been seen on the sun, where that shear plays out between neighboring streams of solar plasma.

Researchers captured the swirls using the U.S. National Science Foundation's Daniel K. Inouye Solar Telescope in Hawaii and then compared the images against computer simulations of the sun's visible surface, called the photosphere. The findings were published Aug. 5 in the journal Nature .

A time-lapse view of the Inouye telescope's obseravtions of the sun. (Image credit: NSF/NSO/AURA/MPS ) The resulting time-lapse footage from the telescope is unlike anything solar physicists have captured before. It shows dozens of tiny vortices clustering along the edges of magnetic regions, some accompanied by delicate, dark stripes known as striations, according to a statement from the National Science Foundation. The telescope's resolution is sharp enough to pick out structures just tens of miles across — a scale so fine that the vortices had gone completely undetected until now.

When the researchers lined the images up against their simulations, the match was close enough ‪—‬ down to the average spacing between vortices ‪—‬ to confirm that they'd finally caught the instability in action.

Beyond the impressive visuals, researchers think these swirls may serve an important function. By constantly twisting and mixing the sun's magnetic-field lines, these vortices may play a part in building up the energy behind solar flares and coronal mass ejections .

RELATED STORIES

'The fate of Earth depends on a delicate balance': Our planet may survive the death of the sun after all, new models hint

Scientists discover the possible origin of the sun's magnetic field, and it's not where they thought it was

How long does it take the sun to rotate?

The effect also might help to explain the long-standing mystery of why the sun's outer atmosphere, or corona, reaches temperatures of around 2 million degrees Fahrenheit (around 1 million degrees Celsius) while its surface remains hundreds of thousands of degrees cooler , Thomas Rimmele , chief technologist at the National Solar Observatory, said in the statement.

The team now plans to use computer programs to automatically track these vortices across future observations. This will help them measure exactly how much these swirls contribute to heating the corona and driving the explosive space weather that can disrupt satellites , communications and power grids on Earth.

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

Full text

The Caspian Sea, the world's largest inland body of water, is in major trouble. Since the mid-1990s, this saline lake, located between Europe and Asia, has lost an area nearly the size of Sicily. But until now, it wasn't clear why.

Researchers finally provided answers in a study published June 5 in the journal Earth's Future . Using ground- and satellite-based observations, they connected falling water levels to a drastic drop in river inputs from Iran, Turkmenistan, Azerbaijan, Kazakhstan and Russia, all of which border the Caspian Sea.

The team particularly highlighted losses from Russia's Volga River, the longest river in Europe, which supplies roughly 80% of the Caspian's inflow.

To find out why the Volga is discharging less water into the Caspian Sea now than it did 30 years ago, we spoke with Amir AghaKouchak , a co-author of the study and professor of civil and environmental engineering at the University of California, Irvine.

AghaKouchak, who directs UC Irvine's Center for Hydrometeorology and Remote Sensing, explained why river inputs have dropped and how decades of water withdrawals have put the Caspian Sea on a path to collapse. He also touched on a solution to this crisis — and revealed why Russia's hunger for ammunition to attack Ukraine is making it harder to save the Caspian Sea.

Sascha Pare: Falling water levels in the Caspian Sea have been attributed to natural variability, and more recently, there's been a focus on climate change and increased evaporation due to rising temperatures. But what did the results of your study show?

Amir AghaKouchak: People quickly jump to conclusions about climate change . Of course, that is important, but it cannot explain the change that we have observed. Over the last 30 years, water levels in the Caspian Sea have shrunk by around 2 meters [6.5 feet] or so. Just to put things in perspective, this translates to about 24,000 square kilometers [9,300 square miles] of sea area loss — about 5%. When we look at drivers, we notice that rainfall hasn't changed significantly over the lake and upstream, so the decline cannot be explained simply by changes in rain.

Evaporation has increased, so it explains maybe about one-third of the change. There is less water in the lake, and that indicates there's probably some water lost along the way. We don't have a lot of human-related data ‪—‬ for example, water withdrawal for irrigation or groundwater wells. That area is very difficult; the countries are not open to sharing data. A lot of our results are based on looking at proxies. Our interpretation is that temperature caused about 30% of the change and the rest is probably because of human activities.

SP: Which human activities does your research suggest are responsible for the Caspian Sea's decline?

AA: Several of the countries on the Caspian Sea have built large dams — Russia, for example. They are all desperate for hydropower. All of the countries around the lake [have also built infrastructure to supply agriculture], and one reason is that they want to create jobs and improve the local economy. Water demand is increasing in all those countries around the Caspian Sea. It's not surprising to see lakes drying or groundwater levels falling.

It's hard to find data about how much water countries are using; they keep data very private, especially places like Russia. We don't have detailed information, but the Volga-Don Canal in Russia is one of the projects that is contributing to this crisis. [ Editor's note: The Volga-Don Canal is a ship canal that connects the Volga and Don rivers at their closest points. ]

Opened in 1952, the Volga-Don Canal connects the Volga, which flows into the Caspian Sea, and the Don, which empties into the Sea of Azov (a northern extension of the Black Sea). (Image credit: Gaika102 via Getty Images) But we cannot blame this whole thing on that. We have to be careful with our language; we can't say one project is responsible. Most likely, this is the collective impact of many different small withdrawals from here and there, overall contributing to a significant decline in the Volga River outflow into the lake. In the paper, we have different rivers, but the Volga dominates because it's the largest river. The Volga has a higher impact than all the other rivers combined. There are lots of demands along that river basin.

SP: Did you expect human activities to play such a big role, or were you surprised?

AA: We did not have any expectations at the beginning. In fact, when we started this project, I reached out to a lot of colleagues from neighboring countries around the Caspian Sea. We didn't want to do it on our own, sitting in California, looking at another area of the world, without having local people involved. If you look at the co-authors, you see people from Azerbaijan, Iran, Kazakhstan — so, we have local experts. We thought this was really important to have those perspectives. I didn't want to reflect the opinions of only one research group. What we have here is the collective opinion of people who have not met in person; we only exchanged emails to look at the results, and then we came up with those conclusions.

We do have a Russian involved, but not from an official Russian institution. I thought this was really important, because there's more back-and-forth [with a large group of people]. If two people have to convince themselves of an idea, it's much easier than with 20. Also, we brought in people from other countries who had projects in the area; for example, there is one group from Finland and another group from Germany. We tried to come up with something that this large group of people could agree on. It was not easy, but it worked out.

The Volga River supplies roughly 80% of the Caspian Sea's inflow, but huge water demands have decreased its flow. (Image credit: ESA, CC BY-SA 3.0 IGO ) SP: What is the worst-case future scenario for the Caspian Sea if shoreline countries fail to decrease their water demands?

AA: My concern is that, in the same general area, we have the Aral Sea, which is now about 5% to 10% of what it used to be , and Lake Urmia, in northwest Iran, that has also shrunk really fast and is now very small, with low levels of ecological activity. So we have two smaller lakes in a very similar climate that, from an ecological perspective, are dead. Because we have seen those degrade, we have to consider that this is the worst-case scenario for the Caspian Sea, too.

The Caspian Sea is much, much bigger. I hope it doesn't happen, but it's on the same path. Again, it's much larger, so it would probably take 100 years [to shrink as much as the Aral Sea and Lake Urmia], even with the current trend. But because we have seen those lakes shrink really fast, it deserves attention. Due to the size of this lake, the potential impacts will be huge. Even if 20% of the Caspian Sea dries out, the consequences will be very significant. It would be a disaster.

SP: You mentioned that drying lakes can become ecologically "dead." What other impacts could the Caspian Sea's falling water levels have on the surrounding landscape and people?

AA: The lake's retreat will expose the lake bed, and because it's saline, you have all kinds of minerals that will be exposed. With the next strong wind, they will go into the air. The health impacts include respiratory diseases, and even eye diseases, that are linked to salt minerals in the atmosphere. Those are the immediate consequences. Additionally, those minerals can travel long distances. In general, salt deposition on agricultural land reduces productivity. It takes a while, but if those lands are exposed for a long time, this will have an impact on the agricultural sector in those areas, which means potential food security issues. We don't exactly know what those materials are on the lake bed, so it's hard to know precisely what the consequences will be.

There is also serious shipping happening in this region. Right now, most of the shrinking is happening in the Caspian Sea's northern part , which is shallower [than the rest of the lake]. The drop in water levels is about 2 meters, and you can imagine that if you have a port and water levels drop more, you won't be able to operate. You'll need to extend your shipping routes southward; you'll need to invest and spend a lot of money to use other ports and docking sites. We have evidence of that around the Aral Sea and also Lake Urmia. There used to be these docking sites and ports, but those are now far away from the water.

Related stories

'It cuts both ways': Positive tipping points can restore wrecked ecosystems — we just need to trigger them, Earth system scientist Tim Lenton says

'In every continent where humans are present, water bankruptcy is manifesting itself': Exiled Iranian scientist Kaveh Madani on our desperate need to preserve our most precious resource

War has brought Iran's water crisis to a breaking point: 'Things will collapse unless there is meaningful structural change'

SP: What must bordering countries do to save the Caspian Sea and prevent these negative effects?

AA: Demand management is absolutely critical. Demand management doesn't get a lot of attention, because unlike big projects that make the news, management means you have to cut things and do things that people don't notice. Some people argue that politicians don't show a lot of interest in demand management, but it's something that we have control over. The technological opportunities are there.

But when you look at that area, there is a big war ‪—‬ so probably it's not the priority to invest in management. There are a lot of things that we don't talk about in the paper that are relevant. I recently learned through discussions with people from the region — this is something that came to us from co-authors and collaborators — that many countries in that area grow cotton, which uses a lot of water. It has a lot of applications, but apparently, one of them is military. Cotton is used to produce ammunition. Russia needs a lot of ammunition, so not only are they growing it themselves, but they also bought neighboring reserves from Uzbekistan and Kazakhstan . So now, even those countries have an incentive to grow more cotton. This is driving more irrigation, more groundwater extraction, more surface water depletion, and probably more impacts on the Caspian Sea.

This interview has been condensed and edited lightly for clarity.

Full text

The U.S. Food and Drug Administration (FDA) has approved a new seasonal flu vaccine that uses mRNA as its base.

Moderna, the vaccine's maker, announced Wednesday (Aug. 5) that a vaccine advisory committee to the FDA unanimously agreed that the shot is safe and effective for preventing illness from seasonal influenza. The shot, called mFlusiva or mRNA-1010, has been approved for people ages 50 and older.

Similar to mRNA vaccines against COVID-19, the vaccine uses the genetic molecule mRNA to encode instructions for cells in the body. In this case, it prompts the cells to build hemagglutinin (HA), a protein found on the surface of influenza viruses. Flu viruses use this protein to infiltrate human cells, and it's the same protein targeted by conventional flu vaccines. Following vaccination with mFlusiva, the immune system learns to recognize the HA protein on targeted flu strains, thereby improving its ability to block the viruses from infecting cells.

The HA protein poses a major challenge for conventional flu vaccines because it mutates very rapidly as the viruses multiply and spread — and these existing vaccines take months to manufacture. That's one reason the resulting shots aren't always a great match for the flu viruses that end up circulating in any given year. A benefit of the mRNA platform is that vaccines can be produced very quickly and thus made to better match the season's dominant strains.

mFlusiva's approval was based in part on a late-stage clinical trial involving over 40,000 people ages 50 and up who were vaccinated for the 2024-2025 Northern Hemisphere flu season. Half received the mRNA-based shot, and half received a standard dose of a conventional flu vaccine (as opposed to the higher-dose option available to older adults in the U.S.).

Of the 40,000 participants, about 970 got sick with flu, including 2% of the mRNA group and 2.8% of the standard group. That means the mRNA vaccine was 26.6% more protective than the conventional vaccine against illness from flu in this trial.

mFlusiva also offered superior protection against each of the three flu subtypes included in the vaccine. And in participants ages 65 and up, the vaccine was 27.4% more protective, overall.

More people in the mRNA group experienced side effects like injection-site pain, headache, fatigue and achiness, compared with the standard group, but these effects were mostly short-lived and mild to moderate in severity. These side effects were consistent with what was seen in earlier trials of the vaccine.

An additional trial involved a group of adults ages 65 and older and directly compared mFlusiva to an existing high-dose flu vaccine, which is approved only for older adults in the U.S. The findings suggested that the mRNA vaccine generated superior immune protection than the conventional option, based on antibody levels.

RELATED STORIES

Nobel Prize in medicine goes to scientists who paved the way for COVID-19 mRNA vaccines

New mRNA vaccine for deadly brain cancer triggers a strong immune response

An experimental mRNA treatment counters immune cell aging in mice

These results led the FDA advisory committee to decide that the vaccine's benefits outweigh its risks, and the agency ultimately approved it. For adults ages 50 to 64, the vaccine earned standard approval, and for adults 65 and up, it earned "accelerated approval," meaning further trials will be conducted to confirm the benefit of the shot over the high-dose option for this population.

"Flu remains a significant public health challenge, and mFLUSIVA provides an important new option for America's seniors," Moderna CEO Stéphane Bancel said in a statement .

The approval appears to mark a shift in the government's stance on mRNA vaccines.

Under the second Trump administration, the federal government has been broadly retracting its support for mRNA vaccines , directly canceling hundreds of millions of dollars in funds for research and development and indirectly sending a chilling effect through the mRNA field, experts previously told Live Science . Even Moderna has been pulling back from investments, citing opposition from the U.S. market . Health Secretary Robert F. Kennedy Jr. has repeatedly voiced his opposition to mRNA technology.

When Moderna initially filed for mFlusiva's approval in February, the FDA declined to review the application — but then changed its mind one week later . The exact motivation for this reversal was unclear to experts .

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

Full text

In the dead of night on Wednesday (Aug. 5), part of a defunct SpaceX rocket slammed into the moon at more than 5,000 mph (8,000 km/h) — and a Korean lunar satellite has just shared the first images of the aftermath.

A series of photos snapped by South Korea's Danuri lunar orbiter on Wednesday reveal a small, blackened crater at the alleged site where a used Falcon 9 upper stage crashed into the lunar surface around 2:35 a.m. ET. The crash occurred near Einstein crater, on the northwestern limb of the moon, close to the terminator zone where lunar day and night meet. The photos were shared by the Korea AeroSpace Administration on social media on Thursday (Aug. 6).

Because this region was on the sunlit side of the moon at the time, the initial flare from the crash was not visible from Earth; however, scientists at the European Southern Observatory's Very Large Telescope in Chile reported seeing evidence of a debris plume lifting off of the lunar surface shortly after the collision, according to the CBC . (At this time, no photos of the debris plume have been released).

A preprint study shared earlier this week predicted that the crash would create a new crater on the moon measuring roughly 90 feet (27 meters) wide and up to 16 feet (5 m) deep — a dent too small to be seen from Earth, even with a good telescope . No measurements of the newfound crater have been shared yet, but they are likely to arrive in the coming days as other lunar orbiters, such as NASA’s Lunar Reconnaissance Orbiter, and Earth-based observatories get a chance to examine the crash site.

What caused the crash?

Images from Korea’s Danuri orbiter showing the before and after views of the lunar crash site. (Image credit: KASA) The pesky piece of space junk that hit the moon on Wednesday is the second (or upper) stage of one of SpaceX's Falcon 9 rockets that launched into space on Jan. 15, 2025. It was carrying a pair of lunar landers: the Blue Ghost lander developed by private company Firefly Aerospace, which successfully touched down on the moon in March 2025; and the Hakuto-R lander developed by Japanese company ispace, which lost contact with Earth and crash-landed on the moon in June 2025.

Unlike Falcon 9's first (or lower) stages, which can return to Earth for reuse, the rocket's second stages are discarded in orbit. This rocket piece spent nearly a year drifting in an elliptical orbit around our planet before a combination of solar radiation and the moon's gravitational pull nudged it away from us, according to Live Science's sister site Space.com . In May this year, independent astronomer Bill Gray first predicted the crash and when and where the collision would occur, before other experts confirmed his calculations.

The 4.4-ton (4 metric tons) rocket part is estimated to have struck the moon at roughly 5,400 mph (8,700 km/h). While many times faster than the speed of sound, that’s still much slower than the average meteoroid strike on the moon, making the crash that much harder to see.

More to come

This is not the first time that human-made objects have impacted the moon, nor will it be the last. Studying these crashes can lead to important insights about the moon’s composition.

For example, the 2009 crash of a Centaur rocket's upper stage near the moon's south pole was analyzed by NASA's Lunar Crater Observation and Sensing Satellite mission, which revealed subsurface ice in the area. As a result, the lunar south pole is now the most coveted location for missions that aim to establish a permanent human presence on the moon.

Studying human-made impacts like these will also be essential for managing the potential risks to future astronauts and missions, like the permanent moon base NASA hopes to build.

"Now that we're in an era where we're going to have more and more astronauts going back to the moon, perhaps permanently, we also want to understand how much of a risk these artificial impacts pose," Benjamin Fernando , a postdoctoral researcher at Los Alamos National Laboratory and co-author of a recent study on the impact, told CBC .

"It presents a different set of hazards than say the Apollo astronauts faced on the lunar surface from the average background meteoroids that are striking the moon every day. And what we want to try and do with this event is understand both the direct [and indirect] hazards."

Full text

Physicists have built a crystal out of an "einstein" tile — an elusive shape mathematicians spent years searching for — and found that it bends light in a way no ordinary crystal can.

In the new study, published July 29 in the journal Nature Communications , researchers arranged this never-repeating tile shape into a pattern of nanoscale holes. This structure, known as a photonic crystal, is designed to control the way light moves through it. When the team shined a laser at it, the crystal produced a swirling, pinwheel-shaped scattering pattern that responded differently depending on which direction the incoming light was spinning.

The tile solves what mathematicians call the "einstein problem," a decades-old puzzle asking whether a single shape could cover a flat surface endlessly without ever repeating its pattern. (The name is a pun on "ein stein," German for "one stone," not a reference to Albert Einstein ).

In the 1970s, mathematician Roger Penrose showed that two different shapes, used together like floor tiles, could cover a surface without the pattern ever repeating. Then, in 2023, geometry enthusiast David Smith and collaborators finally found one: a 13-sided shape that they nicknamed the "Smith hat "

Yuto Moritake , an experimental physicist at the University of Tokyo, first came across the hat tile in a popular science book in 2024. Moritake specializes in photonic crystals — materials patterned with repeating structures (often much thinner than a human hair) that bend and steer light for uses like lasers and optical sensors. Almost all photonic crystals rely on patterns that repeat in a regular grid. Moritake wondered what would happen if he swapped that repetition for the hat tile's never-repeating arrangement.

"I decided to include it in our photonic crystal structure," Moritake told Live Science, describing it as the starting point for the project.

To build the structure, Moritake and his colleagues used two precision manufacturing techniques, called electron beam lithography and etching, to punch hundreds of thousands of tiny holes — each just 100 nanometers in radius, or roughly 500 times thinner than a human hair — into a thin film of silicon nitride. This ceramic material is commonly used in computer chips. The holes were arranged according to the hat tile's pattern, covering a chip roughly half a millimeter across ‪—‬ about the width of a pencil tip.

When the team shined a laser at the finished chip, the light scattered, or diffracted, into a pinwheel-shaped pattern that appeared on a screen. Moritake first photographed the colorful pattern using his iPhone's long-exposure mode. The setting captured faint light over several seconds. Then, he switched to specialized cameras for more precise measurements.

The pattern was made up of well-defined bright spots, called Bragg peaks, that stayed in the same positions no matter where on the chip the laser hit. That consistency confirmed that the structure had the kind of long-range, predictable order found in a quasicrystal — a class of material whose atoms (or, in this case, holes) follow an orderly pattern that never repeats, unlike the repeating grids found in ordinary crystals such as table salt or diamonds.

Related stories

'Crystals' of space-time could be the origins of certain rare black holes, theoretical study hints

Scientists unveil new type of 'time crystal' that defies our traditional understanding of time and motion

Distorted crystals use 'pseudogravity' to bend light like black holes do

Because the hat tile itself has no mirror symmetry — meaning it doesn't look the same as its own reflection, similar to how a left hand differs from a right hand — the scattering pattern it produced was also asymmetrical in this way. Scientists call this property chirality.

That chirality led to the study's biggest surprise. Light can be circularly polarized, meaning it spins either clockwise or counterclockwise as it travels, like a corkscrew. When Moritake tested the structure with both types of spinning light, he found a subtle difference in how each direction scattered off the crystal. Ordinary quasicrystals, which do have mirror symmetry, cannot produce this kind of effect.

"This structure can have some kind of circular polarization dependence," Moritake said, adding that the effect — made only possible by the tile’s asymmetry — was not something he initially expected to find .

Now, Moritake wants to apply this never-repeating pattern to control light traveling inside a photonic chip, rather than light that simply bounces off its surface. This could lead to future uses in optical communications and optical computing ‪—‬ technologies that use light instead of electricity to transmit or process information, Moritake said.

See how much you know about Albert Einstein with our Albert Einstein quiz!

Full text

A German startup has debuted the world's first diamond-based quantum computing system to exceed 10 qubits.

The machine, built by engineers at Saxon Q, is a nitrogen-vacancy (NV) quantum computer, meaning it uses defects in synthetic diamonds as quantum bits (qubits) to perform quantum operations. Qubits can be manipulated to represent the 0s and 1s of data, as well as quantum states that are superpositions of both the 0s and the 1s.

The hardware is currently available in rack-mounted systems featuring up to 128 qubits, with 512-qubit configurations available for delivery next year. According to the company's road map , the goal is to scale to 10,000 qubits and beyond after 2030.

Although the technology existed before this debut, scientists have found it difficult to build systems beyond 10 qubits due to the difficulty of creating nitrogen vacancy qubits.

Although Live Science saw a technical white paper outlining how the technology works, there didn't appear to be any published research demonstrating a functional quantum computer based on the NV architecture operating with more than 10 qubits before this launch. It's still unclear exactly how well the firm's quantum computers stack up against other platforms.

Quantum mechanics and flawed diamonds

In the 1970s, scientists discovered that certain diamonds shone with a brilliant red light when illuminated in a specific way. Subsequent research determined that the optical shift was caused by annealing radiation damage that attracted isolated substitutional nitrogen atoms. In other words, nature occasionally produces a diamond that has a nitrogen atom where a carbon atom should be.

The unintegrated nitrogen atoms are drawn to the vacancy in the otherwise perfectly ordered carbon atoms inside the diamond. Although these flawed diamonds occur rarely in nature, scientists can create diamonds with nitrogen vacancies in laboratories.

Functionally, the single nitrogen atoms inside the vacancies act as if they were "trapped," and their electrons can "spin" independently of the electrons inside the surrounding carbon atoms.

Scientists exploit this "spin" feature by using special lasers to put the nitrogen atom's electrons in a specific state they measure as "zero." They can then use microwave pulses to precisely manipulate the electrons into numerous configurations, including quantum states that binary "bits" can't achieve.

We have a fully functioning quantum computer.

Marius Grundmann, professor of experimental physics at Leipzig University and co-founder of Saxon Q,

Marius Grundmann , a professor of experimental physics at Leipzig University and co-founder of Saxon Q, said the breakthrough that allowed his firm to push past the 10-qubit barrier was a materials discovery.

When the Saxon Q team creates the vacancy inside its lab-grown diamonds, they co-implant sulfur. Per Grundman, “That's the key technology point. Because the sulfur lifts the chemical potential to a point that it's negatively charged. The sulfur supplies the electron; the sulfur also makes the vacancy attached to the nitrogen with a very high yield."

In other words, by co-implanting sulfur atoms, the Saxon Q researchers can express greater control over the individual qubits. Once the qubits are set with lasers and subjected to microwave pulses, they're ready for error correction. In a statement , Saxon Q representatives said the qubits achieved a 99.92% fidelity rate — fewer than one error per 1,000 operations — before error correction.

However, Grundmann told Live Science that the latest numbers as of July 22 showed 99.98% fidelity in single-qubit operations. That's comparable to state-of-the-art error correction results reported by other quantum computing labs, such as those from IBM and MIT , although it wasn't possible to independently verify the results.

Room-temperature quantum computing

It's difficult to compare Saxon Q's diamond-based NV systems with more established quantum computing platforms, such as superconducting qubits. Most recent research on NV systems has focused on quantum sensing applications , though at least one preprint study discusses hybrid NV/superconducting systems.

"We have a fully functioning quantum computer," Grundmann told Live Science. According to Grundmann, the company's quantum computers are on a par with those using other modalities. "We have a quantum computer that can execute quantum code that you can reach via the network, and it is a multiuser, multitasking, multicore system," he said.

If the Saxon Q systems ultimately prove comparable to existing solid-state quantum computing architectures, they'd be among the first generation of room-temperature quantum computers to reach performance similar to systems that require cryogenics and on-site monitoring.

The ease of setup is notable. Saxon Q's devices can be slotted into a standard computer rack and plugged directly into alternating-current power.

Related stories

Scientists trained an AI model using an IBM quantum computer — and it answered questions correctly that the base model couldn't

New chip harnesses quantum computing's biggest weakness — and tries to turn it into a strength

China unveils first-of-its-kind 'dual-core' quantum computer — its makers say it improves stability and efficiency

This provides a clear near-term advantage for clients who want to run algorithms on a quantum system without operating through the cloud. Grundmann said this could be especially important in edge computing scenarios, such as autonomous driving or robotics, where cloud communication could produce unacceptable latency.

Research comparing solid-state quantum systems indicates that superconducting quantum computers would typically operate faster than diamond-based NV systems. It's unclear, however, what the trade-off between processing speed and cloud latency would be or whether it could be addressed through scaling.

The current challenge to scaling diamond-based NV quantum computers beyond the 512-qubit range is the size of the chips. Saxon Q's current chips are limited to supporting either eight or 16 qubits. For more powerful systems, scientists will need to squeeze hundreds ‪—‬ or even thousands ‪—‬ of qubits onto a single array to support operations requiring hundreds of thousands or millions of qubits.

Can you match these ancient devices to their pictures? Find out with our computing quiz!

Full text

A fleet of telescopes around the world just captured the first shocking moments of a supernova explosion in the most detailed-ever observations of their kind.

The explosive event was first spotted by China's Einstein Probe space telescope as a flash of X-rays in March. Ground-based telescopes jumped in within the hour and quickly identified the flash as a supernova, a powerful explosion of a massive dying star. Before long, a big team of observatories was on the job, including the wide-field Vera C. Rubin Observatory , which began its ambitious 10-year survey of the sky in late June.

With this fleet of telescopes, astronomers charted how the light of the supernova, located 500 million light-years away, changed over time. They reported their results in two papers published July 14 in The Astrophysical Journal Letters.

Each science team verified the "shock breakout" ‪—‬ the point when a supernova emits its first light from the explosion as a huge shock wave. The recently detected supernova is only the second stellar explosion in the past 20 years in which a shock wave was observed, because these events tend to be short — only seconds to hours, the researchers noted.

They classified the stellar explosion as a Type Ic broad-lined (Ic-BL) supernova, meaning that the blast included jets with material moving close to light speed. This type of supernova is often associated with gamma-ray bursts, which are extremely powerful flashes of high-energy radiation from space. However, the recently detected supernova presented some surprises: Its shock breakout was the faintest seen in an Ic-BL supernova, and no gamma-ray bursts were observed.

"Follow-up observations using the most sensitive facilities found no evidence" of the gamma-ray bursts, Brendan O'Connor , an astrophysicist at Carnegie Mellon University and co-author on one of the studies, said in a statement . "One possibility is that the jet was 'choked,' either by the surface of the star or by circumstellar material surrounding the star."

Telescopes of the world unite

The Vera C. Rubin Observatory happened to be monitoring the area (left) when the supernova erupted. Rubin and Dark Energy Camera images of the blast are shown in boxes to the right. (Image credit: NSF–DOE Vera C. Rubin Observatory/NOIRLab/SLAC/AURA) The Rubin telescope, located in Chile, happened to spot the explosion because it was viewing that particular patch of the sky at the time, known as the COSMOS Deep Drilling Field.

"Thanks to Rubin's rapid cadence and unprecedented sensitivity, continued observations are expected to provide detailed, long-term records of the supernova as it evolves for years to come," the science teams said in the statement. The Rubin Observatory will spend the next decade mapping the entire southern sky many times over, creating a time-lapse movie of the cosmos expected to reveal millions of supernovas , asteroids, comets, and other transient events.

As Rubin spots events in the sky, the Dark Energy Spectroscopic Instrument on the National Science Foundation's Nicholas U. Mayall 4-meter Telescope at Kitt Peak National Observatory in Arizona is tasked with springing into action for follow-up observations. The instrument confirmed the recently detected explosion as an Ic-BL supernova.

When the scientists dove into 10 years of data from the 570-megapixel Department of Energy Dark Energy Camera on the National Science Foundation's Víctor M. Blanco 4-meter Telescope in Chile, they found a "blue source" just where the supernova later occurred, revealing clues about what the star system and its environment looked like before it exploded.

Meanwhile, a team led by Jillian Rastinejad , an astronomer at the University of Maryland, College Park and co-author of the second study, used the Gemini Multi-Object Spectrographs on two telescopes — Gemini North in Hawaii and Gemini South in Chile — to investigate the explosion, among other observatories. Those telescopes also confirmed the explosion as an Ic-BL supernova, without the jets, and the astronomers obtained archival details on the star's structure and environment before their demise.

Related stories

Scientists mapped the shape of a supernova for the first time ever – and it's not what we expected

Supernova that lit up Earth's skies 843 years ago has a flowering 'zombie star' at its heart

Scientists find evidence of 'supernova graveyard' at the bottom of the sea — and possibly on the surface of the moon

"Our observations allowed us to study the physics of three pieces of this explosion: the X-ray shock breakout, the accompanying supernova, and the interaction of the supernova with material previously cast out by the dying star," Rastinejad explained in the statement. "With this information we were able to map out the structure of the material surrounding the star and understand the star's violent lifestyle before it collapsed."

The star was about 20 times the mass of the sun and was classified as a Wolf-Rayet — a star that used up most of its hydrogen early in its life. Just before the star exploded, it periodically ejected massive chunks of hydrogen and helium, leaving behind only carbon and oxygen. These chunks of material generated shells that were visible in telescopes as individual events, including the first X-ray shock.

"This is the first time we've mapped out the pre-explosion environment of a star that has been stripped of hydrogen and helium," said Gokul Srinivasaragavan , an astronomer on Rastinejad's team who was a doctoral student at the University of Maryland at the time of the study. "Going forward, I'm excited to observe more shock breakout events in similar detail to test if all stripped stars have a similar ‘lifestyle' prior to collapse and what, if any, differences we see."

Full text

Don’t believe the Kremlin. It claims that from 2020 to 2024, the Russian economy grew by 13%. Instead, believe the lights Russians leave on at night. They reveal the country’s GDP actually shrank by 8% over that period. So says Sweden’s foreign minister Maria Malmer Stenergard in a New York Times op-ed .

Stenergard wasn’t entirely guessing. Sweden’s intelligence services arrived at that figure by measuring nighttime luminosity, an increasingly mature and rigorous branch of economic geography.

Because unlike the Kremlin, night lights don’t lie. Artificial light at night ( ALAN ) provides strangely honest data on GDP, poverty, war, and many other things governments would rather scientists didn’t measure. Recent analyses of NASA satellite data have made the field even more relevant, revealing the night side of the planet to be a subtle fact-checker of misinformation.

An accidental science born in the Cold War

Like many useful things, ALAN began as an accident — specifically, a Cold War one.

In the 1960s, the U.S. Air Force launched a constellation of weather satellites under the Defense Meteorological Satellite Program . The satellites were designed to track cloud cover so military planners could decide where to fly bombers and surveillance aircraft. By happy coincidence, the same sensors built to watch clouds also picked up everything else glowing in the dark: gas flares, wildfires, city lights. The entire nocturnal fingerprint of human activity.

Moscow’s Red Square at night. Changes in nighttime luminosity show Russia’s economy shrank by 9% from 2020 to 2024 instead of growing by 13%, as claimed by the Kremlin (to the left on this photo). ( Credit : Soeren Stache/picture alliance via Getty Images)

DMSP remained classified until the 1970s, when its data was made available to the civilian and scientific communities. After the fall of Communism, nighttime luminosity joined GPS on the list of military satellite secrets that became everyday tools. In the mid-1990s, operation of the satellites was transferred to the National Oceanic and Atmospheric Administration (NOAA).

Around the same time, a NOAA researcher named Christopher Elvidge had a — pardon the phrase — lightbulb moment. He noticed that the combined wattage visible in DMSP imagery tracked with a country’s GDP astonishingly well. His logic is refreshingly simple: Economic activity needs energy, and a meaningful share of that energy escapes upward as light. Factories run night shifts, offices stay lit, ports glow around the clock. Each country is unknowingly broadcasting its economic report card straight into low-Earth orbit.

A follow-up paper by Elvidge et al . found a correlation ( R 2 ) of 0.97 between lit areas and GDP across 21 countries. For the non-statisticians among us, that is the kind of near-perfect fit that makes most social scientists want to eat their 70-part questionnaires — along with the clipboard.

Yet, much like the laser, which was dismissed when it was invented in 1960 as “a solution looking for a problem,” nobody knew quite what to do with this data. For nearly two decades, the correspondence between night lights and economic activity just sat there, lost aimlessly in the dark.

From party trick to forecasting instrument

The lights came on in 2012, when economists J. Vernon Henderson, Adam Storeygard, and David Weil published “Measuring Economic Growth from Outer Space” in the American Economic Review . This paper turned Elvidge’s neat party trick into a genuine forecasting and auditing tool.

This comparison of satellite images from 2012 and 2021 show a major increase in nighttime luminosity across India, especially in the country’s northern belt, roughly from Delhi to Kolkata. ( Credit : Reddit/MapPorn)

The timing helped: That same year, NASA’s Visible Infrared Imaging Radiometer Suite ( VIIRS ) replaced the old DMSP satellite sensors with roughly 20 times the resolution, each pixel shrinking from 5 kilometers to 750 meters. With a wink to its famous 1972 “ Blue Marble ” photo of Earth, NASA packaged the new nighttime imagery in a product called “ Black Marble .”

Just like that, a new scientific discipline was born: nighttime light remote sensing ( NTL ). Turning nighttime glow into an actual GDP estimate takes a five-step dance with data:

Acquire it . Thank you, VIIRS.

Clean it up . Strip out the noise, such as moonlight, cloud cover, wildfires, and the aurora. Basically, anything that glows but doesn’t pay taxes.

Measure radiance . Expressed in nanowatts per steradian per square centimeter, a unit that gives you the brightness of any patch of Earth you like. It can be a single VIIRS pixel, a city block, or an entire country.

Apply elasticity . Henderson, Storeygard, and Weil found that a 1% change in luminosity corresponded to a 0.3% change in GDP (in countries with solid statistics, anyway). In emerging economies with shakier official numbers, that figure climbs to roughly 0.65%.

Cross-reference and judge . Compare the satellite’s verdict with whatever official figures exist. In 2022, the World Bank indeed found that a 10% luminosity decline in emerging markets lines up with a GDP contraction of about 6.5%.

Where the numbers don’t line up, you’ve either found a calculation error or a government that’s been doing some creative accounting.

The field’s Hall of Fame (and Shame)

North Korea is NTL’s reluctant poster child. The field’s single-most reproduced image is of the Korean peninsula after dark . South Korea blazes like its own private galaxy, centered on Seoul. Meanwhile, North Korea is almost entirely dark, with one single, defiant pinprick of light in Pyongyang — as if Kim Jong Un insists on sleeping with the hallway lights on.

Nighttime image of the Korean Peninsula, captured by DMSP on September 21, 2003, with political borders overlaid on the raw satellite data. The image gained widespread recognition when it was used in a U.S. Department of Defense news briefing in 2006, following North Korea’s first underground nuclear test. ( Credit : U.S. Department of Defense – public domain).

Myanmar delivered the field’s first genuinely quantifiable gotcha . World Bank data showed the country growing at a blistering 10% a year between 1992 and 2006. Satellite imagery told a different story: close to zero growth, with the economy outright shrinking some years. The military junta had simply cooked the books, and nobody outside the country had any way to check until someone thought to ask the night sky.

China is perhaps the emerging field’s most contested case. Economists at the St. Louis Fed found China’s cumulative GDP growth of roughly 122% between 1992 and 2006 stacked up against a luminosity-implied figure of 57% — a 65-percentage-point overstatement with economic and potentially geopolitical consequences.

A landmark 2018 study by Luis Martinez , a professor of policy at the University of Chicago, found that in democracies , a 10% rise in nighttime light tracks a 2.4% rise in reported GDP — almost exactly what you’d expect. This means democratic books are roughly telling the truth. Conversely, in autocracies , the same 10% light increase comes with a much bigger reported GDP bump. Headline finding: Autocracies pad their GDP figures by 15 to 30% relative to what their light emissions support. That estimate tracks the 21-point gap in Sweden’s read on Russia, by the way.

In 2016, when Indian Prime Minister Modi yanked 86% of the country’s cash out of circulation overnight, rural nighttime light dimmed sharply . Cash-dependent regions were hit hardest, even as official GDP figures showed an implausibly gentle slowdown.

Venezuela offers one of the starkest recent examples of economic collapse. VIIRS data shows luminosity down 30–50% across large parts of the country, broadly matching an independently estimated 75% GDP collapse between 2013 and 2021.

The Black Marble, simultaneously brightening and dimming

NTL keeps getting more interesting. A 2026 analysis of NASA’s Black Marble data , which landed on the cover of Nature , found that global nighttime radiance rose 34% between 2014 and 2022. However, that tidy headline number hid a planet brightening and dimming at the same time, a phenomenon researchers call “bidirectional change.”

While the West Coast metros got uniformly brighter, the night lights in large parts of the eastern half of the U.S. dimmed considerably. ( Credit : NASA Earth Observatory/Michala Garrison)

The U.S. provides a textbook example: West Coast metros grew brighter as their populations swelled, while much of the East Coast dimmed — not from decline exactly, but from efficient LED retrofits and the slow-motion restructuring of older industrial corridors.

Internationally, China and northern India blazed brighter with breakneck urbanization, while Paris, London, and the Netherlands dimmed by 20–33%, mostly courtesy of LED swaps and tighter light-pollution rules.

And in early 2022, Europe’s lights dimmed sharply for an altogether less positive reason: The energy crisis triggered by Russia’s invasion of Ukraine. Per Sweden’s accounting, the same war quietly dimmed Russia’s own economy on the other side of the frontline.

More than a GDP gadget

Previously, we would have intuitively guessed that the Earth’s artificial lights keep getting brighter. The realization that they are simultaneously brightening and dimming adds a subtle layer to NTL and suggests it can do more than just read economic data. Such as:

Mapping poverty . Luminosity tracks household wealth closely enough to produce poverty maps down to individual city districts. Think census work without the inconvenience of field surveys.

Monitoring urbanization . Official city boundaries rarely keep pace with fast-growing informal settlements. NTL tracks the real footprint regardless of what the municipal map says.

Tracking electricity access . Another reliable way to annoy a government. A University of Michigan study discovered that Indian villages officially declared “electrified” sometimes had exactly one streetlight hooked up.

Charting wars . Researchers at Wuhan University found that 83% of Syria’s lights had gone out between the start of its civil war in 2011 and 2015. It is one of the largest luminosity collapses ever recorded. In Aleppo, the hardest-hit city, the figure reached 97%. The data showed a near-linear relationship between a province’s light loss and the number of people who fled it, turning satellite photography into an improvised refugee census.

Spotting dark fleets . A Science Advances study exposed more than 900 Chinese vessels illegally fishing in North Korean waters at night. The fishermen were using floodlights to lure squid to their nets.

Disaster response . After Hurricane Maria knocked out Puerto Rico’s grid in 2017, VIIRS imagery showed the entire island going dark in near real time, a stark visualization of the damage that beat official reports by weeks.

Regional favoritism . In a paper that surely belongs in this field’s Hall of Fame, economists Roland Hodler and Paul A. Raschky found that in many countries, a sitting national leader’s home region systematically emits more light — a proxy for public investment — than comparable regions elsewhere in the same country.

The list keeps growing: cross-checking carbon emissions, verifying whether foreign aid actually reaches the ground, and so on. But none of it works with satellite data only, and none of it is infallible. NTL is an extraordinarily good lie detector. Like any lie detector, it still needs a competent examiner asking the right follow-up questions.

If you don’t know anyone, we’re sure the Swedish foreign minister can hook you up.

Western Europe stands out on this image of the world’s eastern hemisphere as the only major region where nighttime lights have consistently dimmed from 2014 to 2022. ( Credit : NASA Earth Observatory/Michala Garrison)

Strange Maps #12xx

Got a strange map? Let me know at strangemaps@gmail.com .

Follow Strange Maps on X and Facebook .

This article How Earth’s night lights became our most honest economist is featured on Big Think .

Full text

An illustration of an ichthyosaur eating a pterosaur, while a pliosaur lurks beneath. (Image credit: Peter Trusler) An ichthyosaur fossil found in Australia's outback may preserve a rare snapshot of a prehistoric food chain. Around 100 million years ago, the enormous marine reptile swallowed a flying pterosaur before being attacked by a giant ocean predator, new research suggests.

The fossil set, nicknamed "BOB" or "Bag of Bones," belonged to a 20- to 23-foot-long (6 to 7 meters) ichthyosaur , an animal that belonged to a group of dolphin-shaped marine reptiles that lived alongside the dinosaurs during the Mesozoic era , from about 250 million to 90 million years ago . But in a new study, researchers discovered fragments of a pterosaur's jaw — along with fish and cephalopod remains — preserved inside the animal's body cavity.

The finding is the first confirmed evidence that ichthyosaurs ate pterosaurs, an extinct group of flying reptiles that ruled the ancient skies, according to the new study, published July 28 in the journal Gondwana Research .

The fossils may also hint at the ichthyosaur's violent end. Its skeleton — discovered in 2019 in Queensland's outback — contains shattered bones, missing ribs and large crushing injuries. Researchers think the gruesome wounds may have been inflicted by Kronosaurus queenslandicus , a colossal crocodile-like apex predator that lived 120 million to 100 million years ago in the ancient Eromanga Sea in Australia.

A rare prehistoric meal

Fossilized stomach contents are a rare find because they can be preserved only under very specific conditions.

"Rapid burial after death is essential to prevent decay and scattering of remains," study co-author Matt White , a research associate at the University of New England in Australia, told Live Science in an email. "The animal must have died soon after its last meal so stomach acids don't fully digest the prey."

The ichthyosaur fossil set "BOB" gets scanned. (Image credit: Kronosaurus Korner Museum Queensland Australia) Researchers used advanced neutron imaging to peer inside BOB's stomach contents without breaking any of the fossilized formations. The scans revealed hidden bones and helped scientists determine that the remains were likely associated with the ichthyosaur's digestive system.

White said that advances in imaging technology are making it possible to uncover evidence that would previously have remained hidden inside fossils.

"Scanning techniques have also improved, which has enabled us to see inside specimens and reveal what they potentially ate," White said. The methods can also reveal "the shape of various organs" and internal bone structures, he added.

The pterosaur remains were found near the front of the ichthyosaur's body cavity rather than in their expected position as digested prey, which White and his colleagues think may be because the ichthyosaur struck the seafloor headfirst shortly after its last meal, shifting some of its stomach contents forward.

"We believe the animal died in the water column — perhaps from a pliosaur ( Kronosaurus ) attack or natural causes — and then sank," White said. "It impacted the seafloor with its rostrum, and connected soft tissues kept most of the body in sequence."

A graphical abstract of the research showing the complex food web of Cretaceous Australia. (Image credit: White et al. (2026)) Reconstructing an ancient ecosystem

Related stories

Oldest 'fish-lizard' fossils ever found suggest these sea monsters survived the 'Great Dying'

New species of Jurassic 'sword dragon' could help solve an evolutionary mystery

'Ash-winged dawn goddess' is oldest pterosaur ever discovered in North America — and it was small enough to sit 'on your shoulder'

The fossil offers a rare glimpse into the food web of the ancient Eromanga Sea, which covered much of central Australia during the early Cretaceous period around 145 million to 100 million years ago.

Although ichthyosaurs are known to have eaten fish and squidlike animals, evidence that they ate pterosaurs , a common predator of the ancient skies, has been elusive. The new fossil suggests that at least some ichthyosaurs may have fed on flying reptiles that landed on the water or died and floated to the surface.

"Such discoveries yield valuable information about trophic interactions and allow us to reconstruct ancient ecosystems in greater detail," White said.

Full text

A camera drone operator captured footage of a humpback whale giving birth off the north coast of New South Wales in what is thought to be a world first.

Alexander Forrest, a commercial drone pilot for the Organisation for the Rescue and Research of Cetaceans in Australia (ORRCA), was conducting aerial observations of migrating humpback whales ( Megaptera novaeangliae ) on July 28. ORRCA is a nonprofit organization focused on the rescue, conservation and welfare of marine mammals.

During the survey, Forrest noticed a whale swimming slightly behind her pod and going unusually slowly. Then, he saw her give birth.

"It's incredibly rare to see a moment like this in the ocean. It's a very rarely documented encounter," Forrest told Live Science. "It's the first drone video as far as we know of a full humpback whale live birth. It gives critical inside moments in the first moments of a calf's life, and also discovers the behaviour of a new humpback whale mother and her freshly newborn calf."

An ORRCA spokesperson said that this is only the fourth time the phenomenon of a humpback whale giving birth had been filmed, in general. Scientists also have footage of other whale births, including a gray whale ( Eschrichtius robustus ) birth filmed off the coast of California and a sperm whale ( Physeter macrocephalus ) recorded giving birth in the Caribbean .

The expectant humpback whale mother swam behind her pod, eventually giving birth by herself. (Image credit: Alexander Forrest/ORRCA (@forrestfov)) The drone video shows the newborn whale surface from blood-reddened water for its first breath. The mother then turns back to reunite with it. The scene was filmed around 0.4 miles (700 meters) from the shore at Kingscliff in northeastern New South Wales.

The birth is "a symbol of one of the strongest conservation efforts in history, returning humpback whales from near extinction to nearly 50,000 on the New South Wales coast," Forrest said. Gathering data on this humpback whale population is critical to "discovering more about where they birth, what happens when they birth, and what those early moments look like."

Humpback whale calves stay with their mothers for their first 11 months. (Image credit: Alexander Forrest/ORRCA (@forrestfov)) The footage will now be analyzed so that a report on the observation can be peer-reviewed and published in a scientific journal. Notably, the mother whale was trailing behind her pod when she gave birth, a detail biologists will compare with observations of other birthing whales that were accompanied by members of their pod. For instance, the 2023 drone footage of the sperm whale giving birth showed that she had 10 adult females near her that formed a protective circle and took turns lifting the calf up so it could breathe at the water's surface.

Related stories

Kelping is a 'global phenomenon' sweeping the world of humpback whales, scientists say

Humpback whales caught on film for 1st time treating themselves to a full body scrub on the seafloor

'I thought I was dead': Terrifying video of humpback whale gobbling up kayaker explained

Throughout their first weeks, calves stay close to their mothers and learn critical behaviors, such as breathing at the surface.

"Humpback whales seem to have an incredible maternal bond with their offspring," Forrest said. "There is an 11-month period where this calf will be raised and feeding on very nutrient-rich milk until it is a yearling and can fend for itself."

A newborn humpback whale calf weighs 1 to 1.5 tons (900 to 1,360 kilograms) and measures roughly 10 to 16 feet (3 to 5 meters) in length. While feeding on its mother's milk, the calf will gain 100 pounds (45 kg) a day and develop the strength and food storage it needs for its first migration. Humpback whales undertake one of the longest migrations of any mammal, travelling up to 5,000 miles (8,000 kilometers) each way between the cold waters of the poles to warm tropical breeding grounds .

Full text

In the past 20 years, an increasing number of bacterial infections caught by children have been resistant to antibiotics, a new study finds.

The research, published in July in the journal JAMA Pediatrics , looked back at the past two decades, as well as looked forward, to predict how resistance rates might change in the decade to come. By 2035, some critical antibiotics will be met with "steep increases" in resistance among common bacteria, such as Acinetobacter baumannii and Klebsiella , the analysis found.

Antibiotic resistance contributed to 840,000 deaths among children under 5 in 2021, global estimates suggest , so if resistance continues to increase, that number could grow.

"These forecasts are warning scenarios, not inevitable outcomes," said study co-author Dr. Yanhong Jessika Hu , a clinical epidemiologist and public health researcher at Murdoch Children's Research Institute (MCRI) in Australia. "The projections broadly assume that historical patterns continue without a major change in direction," Hu told Live Science in an email.

If significant action is taken, that trajectory could be altered, she said.

"Resistance increased across all regions"

Antibiotic resistance is sometimes called " a silent pandemic ," as it poses a formidable threat without stirring as much concern as other health crises, such as viral outbreaks. Resistance has gained attention in recent years, but the data on how it has affected children is patchy, said study co-author Dr. Penelope Bryant , a clinician-researcher at MCRI, Royal Children's Hospital Melbourne, and the University of Melbourne.

"It's perceived as being a lesser problem in children," Bryant told Live Science. "It's not a theoretical problem; it's real, and it's happening. But I think the visibility is just substantially less than in adults."

That's partly because many resistance-tracking efforts combine adult and pediatric data, and the adult data ends up obscuring trends in children, Hu said. "Adult findings cannot simply be applied to children because children have different infection patterns, antibiotic exposure and treatment options," she noted.

Meanwhile, studies aimed only at children tend to vary in the age groups and clinical settings they include, making direct comparisons difficult, Hu said. This lack of granularity hinders efforts to counter resistant infections in children through changes in policy or clinical protocols. Children also have fewer antibiotic options than adults do at baseline, which makes preserving those options even more important, Bryant added.

The analysis found that resistance rates were highest in ICUs, in children under 3 with the most severe infections, and in resource-limited settings. (Image credit: Jacob Wackerhausen via Getty Images) To help fill the gap, the researchers turned to the Antimicrobial Testing Leadership and Surveillance database , which contains clinical samples from children and teens ages 0 to 18 across 82 countries. These bacterial samples were collected between 2004 and 2022 from a variety of medical settings, including hospital wards, intensive care units and outpatient clinics. The included countries ranged from high-income, such as the U.K. and the U.S., to low-income, such as Malawi and Uganda.

The analysis included over 106,000 bacterial samples and assessed which antibiotics these germs were resistant to. The antibiotics were grouped based on the World Health Organization's Access, Watch, Reserve (AWaRe) system : "Access" antibiotics target few bacteria and are recommended as the first choice for most bacterial infections, "Watch" antibiotics have broader effects and should be used selectively for more-resistant infections, and "Reserve" antibiotics are considered last-resort options to use when all alternatives have failed.

"Resistance increased across all regions, showing that this is not only a problem in lower-income countries," Hu said. "However, the highest resistance levels and some of the steepest increases were seen in lower-resource settings."

A wake-up call

Part of the analysis zoomed in on eight bacteria that are considered "critical" or "high" priority by the WHO , meaning they impose a particularly high healthcare burden due to their resistance. Across these germs, resistance rates were similar across high- and low-income countries in 2004, but those trends diverged by 2022.

Overall, resistance to Access antibiotics declined across the study period, but that improvement wasn't the same across all settings. In high-resource countries, resistance to Access drugs fell from 48% to 29% among the eight bacteria, while it rose slightly — from 44% to 47% — in low-income settings. These percentages correspond to the number of bacterial samples from each year that showed resistance to at least one Access drug.

The decline seen in some countries is "almost certainly because of the large investment that high-income countries have put into controlling MRSA," Bryant said, referencing methicillin-resistant Staphylococcus aureus . MRSA can cause painful skin infections as well as life-threatening complications, like sepsis.

With low-income countries, the problem is amplified because of poor sanitation and because of unregulated access to antibiotics.

Dr. Penelope Bryant, clinician-researcher at MCRI, Royal Children's Hospital Melbourne, and the University of Melbourne

Modeling future trends, the team predicts that resistance to Access antibiotics will stabilize or decline across most regions by 2035. However, resistance against Watch antibiotics rose between 2004 and 2022 and is projected to continue increasing. Among the eight critical and high-priority bacteria, it rose from 14% to 24% in high-income countries and from 16% to 48% in low-income nations.

Resistance to Reserve drugs also increased among these germs, rising from 9% to 25% in high-income countries and from 4% to 37% in low-income ones.

While the model predicted "steep increases" in Watch resistance, it projected slower increases in Reserve resistance over the next decade, by comparison. These trends were driven largely by increases in low-income settings, but high-income settings were not spared.

"With low-income countries, the problem is amplified because of poor sanitation and because of unregulated access to antibiotics," Bryant noted. Easy access to over-the-counter antibiotics makes it more likely for the medicines to be misused, which ultimately pressures more bacteria to evolve resistance.

Related stories

Japan's bold experiment to curb antibiotic misuse has been a huge success. Could it work in the US?

'800 seconds for a sick visit': Some factors driving antibiotic resistance have nothing to do with biology, says medical sociologist Julia Szymczak

'They didn't question it': Why doctors prescribe too many antibiotics

This analysis gives a snapshot of how antibiotic resistance is affecting kids now and how it might do so in the future. However, the data it's based on has some limitations. For instance, the samples collected from each country aren't necessarily representative of the entire nation; they came from only certain participating medical centers and each country contributed different numbers and types of samples. So "the results should therefore not be interpreted as precise national prevalence estimates," Hu cautioned.

"But it's the first large dataset, really, to look specifically at children," Bryant said. And "it gives us an idea that things are getting worse."

The team has made their analysis freely available to other researchers on the website AMR in Kids , where "AMR" stands for antimicrobial resistance. Future steps include surveying resistant infections in children in a more representative way, as well as modeling how different intervention strategies could help drive down the rates .

"I'm hopeful that this is a little bit of a sort of wake-up call around children, that it really isn't a problem just in adults," Bryant said.

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

Full text

The patient: A 60-year-old man in Seattle

The symptoms: The man experienced paranoia, as well as auditory and visual hallucinations. He was also very thirsty but was fearful of drinking water that others offered to him.

What happened next: The man went to an emergency department because he was worried that his neighbor poisoned him. Doctors checked his vital signs and conducted a neurological examination, which did not highlight any immediate concerns.

That changed when blood tests revealed that his level of chloride ions — charged particles of the element chlorine — were above the normal range. This result suggested that the patient may have been suffering from hyperchloremia , a condition in which chloride ions build up in the blood. This usually occurs in people who are dehydrated or consume too much salt.

However, when doctors measured the concentration of other ions in the patient's blood, they realized that this initial interpretation was wrong.

In patients with hyperchloremia, the body cancels out the increase in negatively charged chloride ions by depleting negatively charged bicarbonate ions, which the body makes itself. In essence, the body tries to regulate the electrical balance of the blood.

However, in the man's case, tests revealed that his blood was too negatively charged overall. The most likely explanation was that a negatively charged poison had built up too fast for his body to buffer the change by depleting other ions.

The diagnosis: The man's doctors consulted the poison control department and figured out that he was experiencing a case of pseudohyperchloremia, a condition in which the lab test for chloride erroneously detects another negatively charged ion. This makes it seem like a patient's chloride is too high while masking the actual problem.

A common culprit behind this phenomenon is a high dose of salicylate (aspirin) , but the patient didn't have high levels of this pain reliever in his blood.

The next suspect, and the more unusual one, was the poison bromide, a charged form of bromine. Bromine sits just below chlorine on the periodic table, meaning their ions share similar chemical properties, such as charge and size, and bromide has triggered the lab test for chloride in previous reports .

When the doctors measured the patient's bromide levels, they found that they were roughly 230 times higher than the normal range. They diagnosed him with bromide poisoning, also called bromism, which can cause neurological symptoms such as hallucinations and paranoia.

Taking high doses of aspirin can sometimes lead to pseudohyperchloremia, because lab tests for chloride detect the aspirin in a person's blood. In this case, though, aspirin was not the cause. (Image credit: boonchai wedmakawand viva Getty Images) The treatment: The patient remained in a hospital bed for three weeks so doctors could monitor and replenish his electrolytes. On account of his paranoia, he tried to escape from the hospital, so the doctors had to hold him there involuntarily. They gave him daily doses of risperidone, an antipsychotic medication.

His symptoms improved, and he was weaned off medication and ultimately discharged. His condition remained stable during a follow-up assessment two weeks later.

What makes the case unique: After his condition improved, the patient informed the doctors that he had searched for a healthier alternative to table salt (sodium chloride) and had consulted ChatGPT for advice. The chatbot suggested he swap sodium chloride with sodium bromide, and he experimented with that suggestion for three months.

Medicines containing bromide salts used to be popular for treating various conditions, such as headache or upset stomach . These medications fell out of common use when doctors realized that bromide accounted for 8% of admissions to psychiatric hospitals in the early 20th century.

Because it is similar in size and charge to chloride, bromide can fill in for chloride's roles in the body — and that's a problem. For example, chloride contributes to the brain's electrical activity by passing through chloride channels in neurons. Bromide can also move through these channels and thus disrupt normal brain activity, bringing about neurological symptoms .

Bromide poisoning is rare nowadays, so doctors don't typically assume that their patients experiencing paranoia or hallucinations could have this form of poisoning. However, bromide supplements can be easily purchased online, opening the door to such cases, the doctors wrote in a report of the case .

The man's doctors wanted to see whether ChatGPT would again point to bromide salts if asked the same question. They prompted the chatbot with queries about ways to remove chloride ions from the diet, and once again, the bot suggested bromide salt as an alternative. This response didn't provide a health warning about the dangers of consuming too much bromide.

Other dilemmas

6-year-old develops life-threatening allergic reaction after blood transfusion

Junk-food diet caused a teen's permanent blindness

Rare 'spinal stroke' struck an 8-year-old and temporarily paralyzed her

Given the willingness of chatbots to recommend bromide salts, the doctors argued in the report that it would be prudent to more seriously consider the possibility of bromism during psychological evaluations. This case highlights the dangers of relying on AI chatbots for medical advice , the doctors concluded.

Live Science contacted OpenAI, the developer of ChatGPT, about this case and is awaiting comment. According to the service terms at OpenAI, "You should always verify the information provided by ChatGPT for Healthcare and exercise independent professional judgment in decision-making about a patient without relying primarily or solely on the output."

( Live Science first covered this case when it was published in 2025, and at that time, an OpenAI spokesperson said that the company's safety teams were working to reduce the risk of using its services and to train its products to prompt users to seek professional advice.)

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

Can you guess the diagnosis in these strange medical cases? Find out with our diagnostic dilemma quiz!