The findings, published Sept. 21 in The Astrophysical Journal Letters, could change our understanding of what kinds of exoplanets orbit certain stars.

In about 5 billion years, the sun will balloon outward as a red giant after exhausting all of the hydrogen in its core. During this red giant phase, our star will swell to 100 times its present-day size, gobbling up Mercury, Venus and possibly even Earth in the process, Nathan Kaib, a planetary scientist at the University of Oklahoma who wasn't involved in the new research, told Live Science in an email. Following this violent swelling, the sun will crumple into a collapsed core called a white dwarf.

Until now, scientists thought the four outer planets would escape this phase relatively unscathed. A 1998 simulation-based study suggested that the gas the sun spews from its outer layers will push this planetary quartet outward, doubling the radius of each planet's orbit. Only the gravitational impact of a runaway star could destabilize this new order, and such events are thought to occur only once every billion billion years.

Solar recoil

These predictions hinged on the assumption that the sun will eject its outer layers smoothly. However, recent observations of stars suggest a jerkier loss of mass, with gas clouds lofted by plumes of plasma in the sun — as though our star were spitting out mass blobs. "They're kind of like coronal mass ejections, except much larger," Jim Fuller, a professor of theoretical astrophysics at Caltech and co-author of the new study, told Live Science in an email.

Each "blob" pushes the star backward, similar to how a fired bullet causes a gun to recoil. Although every push is small, moving the sun just a few feet per second, many pushes can, over time, impart the equivalent of an enormous kick that alters the sun's speed by several hundred feet per second. In fact, astronomers have already seen white dwarfs suffer such kicks in other star systems.

When Fuller, who studied these kicks, described them to Konstantin Batygin, a professor of planetary science at Caltech and the study's first author, Batygin "realized this would have important consequences for the evolution of our own solar system," Fuller said. That's because the kicks change the sun's gravitational pulls on everything around it.

Chaos at the edge of town

To determine how the sun's mass loss would influence the giant planets, Batygin created computer-based simulations of the sun and the outer planets. The researchers considered three scenarios: if the sun lost its mass smoothly, if it spit out distinct mass blobs during its entire time evolving as a red giant, and if the sun spewed the blobs only during its last stages. The researchers also varied each ejected blob's mass.

The first scenario produced results similar to those of previous studies, with the outer solar system retaining almost all of its features for billions of years. In contrast, the other two cases had drastically different results. Using 672 simulations, the researchers found that if the sun lost mass in blobs, the giant planets would be thrown into chaos, switching places, developing weirdly oval-shaped orbits, and even getting ejected out of the solar system entirely. Most dramatically, in 96% of the simulations with the heaviest ejected mass packets, the planets get destroyed as they crashed into each other.

These developments also happened much more rapidly — less than a few hundred million years after the sun transformed into a white dwarf, or even during the red giant phase, in the most extreme scenarios.

The study's findings hint that analogues of the present-day outer solar system around white dwarf stars are less likely to exist than previously thought. Kaib noted that the study's results suggest that planets with oval-shaped orbits should dominate instead. It's thought that habitable planets could exist around white dwarfs, but this study's findings show that our understanding of those distant star systems may be incomplete, and that such worlds may, in reality, be unstable.

See how well you know our planetary neighborhood with our solar system quiz!

'> Jupiter, Saturn, Uranus and Neptune could destroy each other sooner than we thought, simulations hint

In the wake of this study, Live Science recently polled its readers on whether they felt AI was making them less motivated after using it.

More than 140 readers voted in the poll, with 30% of voters picking "No, I don't use AI at all, so this situation doesn't apply to me."

"If AI were only a tool, controlled by those who use it I might give it a try," said one commenter in response. "But it isn't under control at all. It gets into your bank accounts and moves money without authorization. It encourages suicide and self-harm. It is addictive and makes people lazy. I know someone who uses it to answer questions and while it is detailed, it makes her lazier, asking one question and getting the answer without her searching for it. AI Is a danger. It should not be allowed."

Other comments reflected this sentiment, with one reader writing, "Use of all automatic aids decrease[s] attention and retention. I see it in the use of navigation devices where drivers ought to recognize routes after a couple of times. AI helps research and cuts down lengthy searches, but over use of summaries weakens critical faculties."

Other commenters questioned whether AI is the problem, or if people are just inclined to take the easy option, asking "Would you get the same results if the answers to all but a few problems were provided on separate sheet of paper?" and another writing, "I think researchers should also investigate whether shovels decrease motivation to dig holes with one's hands. AI, like a shovel, is only a tool."

Do you agree with these readers? Let us know in the comments!

'> 'It isn't under control at all': Live Science readers respond to whether AI is making them lazy

"As clinicians, we know that patients lie to us sometimes," Dr. Jeffrey Linder, a general internist and primary care clinician-investigator at Northwestern Medicine, wrote in a commentary about the survey findings.

However, the results are still important because the inappropriate use of antibiotics can contribute to the mounting problem of antibiotic resistance, in which bacteria build up a tolerance to the drugs designed to kill them. Recent data released by the Centers for Disease Control and Prevention (CDC) highlight the extent of the problem: In 2023, drug-resistant bacteria and fungi caused about 9.7 million infections in the U.S., up from 2.8 million in 2017. In turn, the estimated number of deaths caused by drug-resistant germs rose from 35,000 to 50,800 between the two years.

A major factor driving resistance among bacteria is the misuse of antibiotics — in other words, using antibiotics for ailments like viral infections, which the drugs do not treat. For individual patients, such prescriptions don't cure their illness and can result in an upset stomach or issues with the gut microbiome, the community of microbes in the intestines. Simultaneously, they place extra pressure on bacteria to evolve resistance, granting the germs opportunities to pick up new strategies for thwarting antibiotics.

Doctors play a big role in ensuring that the antibiotic prescriptions they write are necessary. Part of their decision-making process comes down to what patients say about their symptoms, some of which may be fabricated, the two new surveys suggest.

Motivations for lying

The new surveys were published Friday (Oct. 30) in the journal JAMA Network Open.

The first survey was completed by nearly 960 adults from across the U.S. in January and February 2026. It asked whether the respondents had ever told a healthcare provider that their symptoms were worse or more long-lasting than they were in order to get antibiotics, as well as whether they'd ever made up symptoms for the same reason. It then inquired about what motivated a given patient to lie — or avoid lying, if they were truthful — as well as posed contextual questions about their demographics, health behaviors and beliefs.

Overall, about 28% of the respondents reported misrepresenting their symptoms, including 26% who exaggerated their existing symptoms and 17% who fabricated symptoms. They most frequently lied about cold-like symptoms, such as fever, cough and sore throat. About two-thirds of these patients were at least moderately confident that their lie would motivate a clinician to prescribe them antibiotics, and 77% said the strategy worked for them at least once.

Often, patients' motivations for lying centered around feeling very sick, holding strong beliefs that antibiotics would help them recover quickly and get back to their responsibilities, and wanting control over the treatments prescribed to them. Compared with people who didn't lie, those who lied were also more likely to feel that clinicians didn't take them seriously or more likely to endorse statements suggesting that antibiotic resistance is a hoax.

"Greater antibiotic knowledge and aversion to taking antibiotics were associated with lower odds of misrepresentation," meaning lying about symptoms, the study authors noted.

Less than 30% of these patients regretted their lies or taking the antibiotics prescribed to them. Those that did express regret noted experiencing side effects, noticing that the treatment didn't help or feeling that the lying itself was morally wrong.

The second survey was much smaller; it included 236 parents and other guardians of children and teens under 18 years old. It posed similar questions but asked whether the caregivers had lied about their children's symptoms and, if so, what motivated them to lie.

About 27% reported lying about a child's symptoms to get antibiotics, with 87% of these caregivers believing that their lying worked at least once. Their reasons for lying were very similar to those expressed by the respondents in the first survey.

"These findings suggest that symptom misrepresentation may contribute to antibiotic overuse in pediatric care," the study authors concluded.

The surveys reached a diverse group of people, including a range of ages, demographics and locations within the United States; however, the respondents were not necessarily representative of the whole U.S. population, the authors noted. The prevalence of lying should therefore be seen as a tentative measure to be checked in future studies, rather than an unbiased, population-level metric, they wrote. It's also possible that not all the respondents who lied disclosed that they did so or remembered that they had done so.

Doctors are still the gatekeepers

Especially in outpatient settings, which account for most antibiotic use among U.S. patients, it can be difficult for doctors to quickly determine whether a patient needs an antibiotic.

One factor is that there's a lack of fast, surefire diagnostic tests for many bacterial infections. Outpatient appointments are also extremely short, so doctors sometimes worry about overlooking an infection that could then get worse, Julia Szymczak, a medical sociologist at the University of Utah School of Medicine, previously told Live Science. Another major factor is that clinicians can face pressure to prescribe antibiotics from their adult patients or from children's caregivers, as the new surveys reflect.

But in the face of these pressures and patients' occasional lies, Linder argues that it's still up to the clinicians to do the right thing. "While clinicians might interpret patients' symptom maximization and questions as demands for antibiotics, symptoms do not determine the indication, a diagnosis does," he wrote in the commentary.

"It does not track that approximately 80% of respondents who reported misrepresenting their symptoms obtained a prescription," he added, "unless prescribers were being willfully deceived or, more likely, just getting worn down with what they perceive as patient demand, regardless of actual clinical appropriateness."

When it comes to making decisions about individual patients, Linder argued that doctors don't necessarily need to think about the big picture of antibiotic resistance and how their patient's case fits into that global trend. Rather, providers should simply ask whether a given prescription is more likely to help or hurt that patient, he said.

"Clinicians' obligation is to act in patients' best interest," he said, "even if patients are hoping we do something different, even if patients do not like it, or even if patients lie about their symptoms."

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

'> Nearly 30% of US adults admit to lying about or exaggerating their symptoms to get antibiotics from doctors, survey finds

Whether AI could ever achieve this level of sustained self-improvement remains an open question. Previous experiments have, however, demonstrated more limited forms of RSI.

In one study earlier this year, an AI system called A-Evolve-Training was tasked with improving the performance of a 30 billion-parameter large language model (LLM).

Over four rounds, A-Evolve-Training modestly improved the LLM's benchmark performance by testing different training techniques and using the results to change how it approached subsequent experiments.

Peter Bentley, a professor of digital biology at University College London, questioned whether such experiments offer a credible route to RSI.

"This idea that some AI is magically going to start generating better and better AIs by themselves [is] nonsense," Bentley told Live Science in an email.

Bentley pointed out that self-improving computer systems have been around for decades. Evolutionary computation and reinforcement learning, for instance, are both based on the principle of systems refining their performance over time in response to feedback.

The much greater challenge, he said, is developing AI systems capable of autonomously conducting the research needed to create more powerful successors — not to mention overcoming their dependence on human expertise and physical computing infrastructure.

"You can't 'recursively self-improve' unless you can also enhance your own training, enhance your hardware, and test it exhaustively in real world settings," Bentley added.

The challenge of true RSI

The study's authors acknowledged some of these challenges. Developing RSI-capable systems could require enormous investments in computing power, as well as human supervision to ensure successive improvements delivered meaningful gains. All of this becomes effectively redundant if the resources needed to develop and test subsequent AI models outweigh any benefits they offer, the authors noted.

Scientists would also need a reliable way to determine whether an AI system was genuinely becoming more capable over iterations, rather than exploiting weaknesses in tests used to measure its capabilities. This is a known problem in AI development, where models can achieve higher benchmark scores by exploiting testing methods or cherry-picking results.

Kayla Blomquist, an AI researcher at the University of Oxford and co-founder of the Oxford China Policy Lab, said that, while the paper doesn't provide a practical road map to RSI, it could help developers and policymakers understand the technical challenges and safety risks associated with increasingly autonomous AI systems.

She also questioned whether there would ever be a clear point at which AI became completely independent of human development. "The paper describes a gradual handover of responsibility rather than a single moment, so there may never be an obvious 'last' model built by humans," Blomquist added.

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

'> Could AI ever build itself? Scientists map the road to the 'last AI built by humans.'

"My first reaction was one of excited disbelief," Matthew Holman, an astrophysicist at the Smithsonian Astrophysical Observatory and member of the Rubin team who wrote the code that helped discover the prospective dwarf planet, said in a statement. "Objects like this are very rare, so I immediately started checking whether I or the code had done something wrong."

Hidden among stars

The potential dwarf planet, designated 2026 ED138, orbits the sun at roughly 46 times the distance between Earth and the sun. The potential new planetoid has been hiding behind galactic camouflage; with its orbit tilted at a 25-degree angle from the solar system plane, it's at just the right angle to get lost in the light from the inner Milky Way, the researchers said.

Many telescopes struggle to see past the millions of stars making up our galaxy's bulge. The Vera C. Rubin Observatory, however, images the entire southern sky every few nights with the largest digital camera ever built, sending out millions of nightly alerts for every object that has changed brightness or position. This makes it exceptional at identifying anything that moves, from solar system objects to transient events like supernova explosions.

The observatory had only just begun its 10-year mission of scanning the sky in July, and ED138 stood out in the data.

Holman works on the software algorithms that process the extreme amounts of data the observatory collects. The code he developed with his colleague, Kevin Napier, compares billions of observations to determine which ones are detections of the same moving object. Running data from Rubin through the code revealed the unusually large object.

By looking at past data from Rubin and archival images from the Dark Energy Camera mounted on the Victor M. Blanco 4-meter Telescope at Cerro Tololo Inter-American Observatory (both in Chile), the team was able to map out ED138's orbit over the past 10 years.

Going beyond Neptune

The Kuiper Belt is a region of the solar system beyond Neptune that is a wild frontier of small, mostly icy bodies made from frozen methane, ammonia, water and rock. It ranges from roughly 30 to 50 times the distance between Earth and the sun, and is home to most of the known dwarf planets, including Pluto. Thousands of Kuiper Belt objects have already been discovered. These space rocks can be frozen time capsules that hold clues about the early solar system's formation.

Dwarf planets are the largest Kuiper Belt objects that we know of. Bigger than asteroids but not quite planets, dwarf planets are big enough to be rounded under the force of their own gravity but too small to clear their orbits like planets do. While the International Astronomical Union (IAU) has officially confirmed five dwarf planets — Pluto, Ceres, Eris, Haumea and Makemake — astronomers studying dwarf planets often recognize at least nine, with several more candidates whose properties are still unknown.

Depending on which astronomer you ask, there could be anywhere from a handful to hundreds of dwarf planets in this part of the solar system. Identifying which Kuiper Belt objects are dwarf planets is tricky because they are often very dim and move very slowly from Earth's perspective, since their orbits are so large and distant. The likeliest candidates have high albedo (surface reflectivity) and diameters greater than roughly 370 to 500 miles (600 to 800 km).

What makes ED138 special is its size, inferred from its brightness, albedo and distance from the sun. It's estimated to have a diameter between 370 to 930 miles (600 to 1,500 km). For comparison, Pluto — the largest dwarf planet recognized by the IAU — is 1,477 miles (2,377 km) in diameter. Ceres, the smallest IAU-recognized dwarf planet, is 595 miles (958 km) in diameter. Future observations of ED138 will give researchers a more accurate idea of its diameter and density to determine if it's round enough to join the dwarf planet club.

"If confirmed as a dwarf planet, ED138 would join a very small family of worlds," Holman said in the statement. "Rubin will image the whole southern sky again and again for ten years, so any others hiding in the places earlier surveys skipped are now within reach."

'> Bright object beyond Pluto may be the solar system's sixth dwarf planet