SoylentNews

Arthur T Knackerbracket writes:

Suleyman says lawyers, accountants, and marketers could be at risk:

Another big name in the AI industry has given an ominous warning about the technology replacing white-collar jobs. This time, the timeline for the automation apocalypse is a lot closer: Mustafa Suleyman, Microsoft's AI chief, thinks AI will replace most white-collar jobs within the next 12 to 18 months.

Speaking in an interview with the Financial Times, Suleyman talked about "professional-grade AGI" and how Microsoft expects it to capture a large share of the enterprise market.

He claimed that this AI model will be able to do almost everything a human professional does. adding that it will allow Microsoft to offer powerful AI tools to clients that can automate routine tasks for knowledge workers.

Suleyman believes that the impact on the global workforce will be immense. He said that almost everyone whose job involves using a computer could be at risk, including lawyers, accountants, project managers, and marketers.

Suleyman believes these jobs won't be at risk within the next five years – a prediction made by Anthropic CEO Dario Amodei in 2025 – but within the next 12 to 18 months.

The Microsoft exec added that in the next two or three years, AI agents will be able to handle workflows of large, complex organizations more efficiently – an area where they still struggle. He also noted that as AI advances, it will become easier to create new models designed for specific needs.

"Creating a new model will be as simple as making a podcast or writing a blog. In the future, it will be possible to design AI tailored to the needs of every institution and individual on Earth," he said.

When Amodei made his prediction that AI could erase half of all entry-level white-collar jobs within five years, he said it would lead to employment spikes of up to 20%.

upstart writes:

Senator: ICE and CBP "have built an arsenal of surveillance technologies":

A few Senate Democrats introduced a bill called the ''ICE Out of Our Faces Act," which would ban Immigration and Customs Enforcement (ICE) and Customs and Border Protection (CBP) from using facial recognition technology.

The bill [PDF] would make it "unlawful for any covered immigration officer to acquire, possess, access, or use in the United States—(1) any biometric surveillance system; or (2) information derived from a biometric surveillance system operated by another entity." All data collected from such systems in the past would have to be deleted. The proposed ban extends beyond facial recognition to cover other biometric surveillance technologies, such as voice recognition.

The proposed ban would prohibit the federal government from using data from biometric surveillance systems in court cases or investigations. Individuals would have a right to sue the federal government for financial damages after violations, and state attorneys general would be able to bring suits on behalf of residents.

The bill was submitted yesterday by Sen. Edward J. Markey (D-Mass.), who held a press conference [video not reviewed -Ed] about the proposal with Sen. Jeff Merkley (D-Ore.), and US Rep. Pramila Jayapal (D-Wash.). The Senate bill is also cosponsored by Sens. Ron Wyden (D-Ore.), Angela Alsobrooks (D-Md.), and Bernie Sanders (I-Vt.).

"This is a dangerous moment for America," Markey said at the press conference, saying that ICE and CBP "have built an arsenal of surveillance technologies that are designed to track and to monitor and to target individual people, both citizens and non-citizens alike. Facial recognition technology sits at the center of a digital dragnet that has been created in our nation."

upstart writes:

AI vision systems can be very literal readers:

Indirect prompt injection occurs when a bot takes input data and interprets it as a command. We've seen this problem numerous times when AI bots were fed prompts via web pages or PDFs they read. Now, academics have shown that self-driving cars and autonomous drones will follow illicit instructions that have been written onto road signs.

In a new class of attack on AI systems, troublemakers can carry out these environmental indirect prompt injection attacks to hijack decision-making processes.

Potential consequences include self-driving cars proceeding through crosswalks, even if a person was crossing, or tricking drones that are programmed to follow police cars into following a different vehicle entirely.

The researchers at the University of California, Santa Cruz, and Johns Hopkins showed that, in simulated trials, AI systems and the large vision language models (LVLMs) underpinning them would reliably follow instructions if displayed on signs held up in their camera's view.

They used AI to tweak the commands displayed on the signs, such as "proceed" and "turn left," to maximize the probability of the AI system registering it as a command, and achieved success in multiple languages.

Commands in Chinese, English, Spanish, and Spanglish (a mix of Spanish and English words) all seemed to work.

As well as tweaking the prompt itself, the researchers used AI to change how the text appeared – fonts, colors, and placement of the signs were all manipulated for maximum efficacy.

The team behind it named their methods CHAI, an acronym for "command hijacking against embodied AI."

While developing CHAI, they found that the prompt itself had the biggest impact on success, but the way in which it appeared on the sign could also make or break an attack, although it is not clear why.

Arthur T Knackerbracket writes:

Why are criminals stealing used cooking oil from Scotland's chip shops?

Police Scotland says organised crime gangs are targeting chip shops, takeaways and restaurants for their used cooking oil.

The liquid is often left in containers outside premises to be taken away to be recycled for potential use as biodiesel, a renewable fuel for transport such as buses and tractors.

Across Scotland, 178 incidents of cooking oil thefts were reported to police between April and October last year.

Grant Cranston said he was surprised by how brazen the thieves who targeted his Inverness chip shop were, adding: "It was broad daylight. There were people walking around."

About 70% of biodiesel produced in the UK is made from used cooking oil, according to UK government statistics.

Prices paid to caterers for their oil can depend on how much is available for collection and its quality, but according to the industry, a restaurant could get about 30p a litre.

On average, thefts of used cooking oil costs the UK Treasury £25m-a-year in lost duty.

Thefts have previously been reported elsewhere in the UK, including in Derbyshire and Gloucestershire.

Police Scotland said the incidents it recorded last year totalled about £20,000 in lost revenue to catering businesses.

Arthur T Knackerbracket writes:

John Carmack proposes fiber-optic loops as high-speed AI cache

Light, not silicon, could someday define how artificial intelligence stores and recalls its knowledge. That's the idea that recently surfaced when John Carmack – the engineer known for his work on Doom and Meta's virtual reality projects – proposed using fiber-optic loops as a form of high-speed data cache for AI models. His brief post on X turned into a dense technical conversation among researchers and technologists intrigued by the blend of classic computing theory and modern optical networking.

The thought experiment began with a number. Single-mode fiber optics can now transmit data at 256 terabits per second over 200 kilometers. Based on that capacity, Carmack estimated that about 32 gigabytes of information are stored in the cable at any given moment.

Light, not silicon, could someday define how artificial intelligence stores and recalls its knowledge. That's the idea that recently surfaced when John Carmack – the engineer known for his work on Doom and Meta's virtual reality projects – proposed using fiber-optic loops as a form of high-speed data cache for AI models. His brief post on X turned into a dense technical conversation among researchers and technologists intrigued by the blend of classic computing theory and modern optical networking.

The thought experiment began with a number. Single-mode fiber optics can now transmit data at 256 terabits per second over 200 kilometers. Based on that capacity, Carmack estimated that about 32 gigabytes of information are stored in the cable at any given moment.

upstart writes:

Europe's $24 Trillion Breakup With Visa and Mastercard:

There's a reason every decentralized system eventually finds its way onto a platform: platforms solve real-world problems that platform users struggle to solve for themselves.

Everyone needs platforms: writers, social media users, people looking for a romantic partner. What's more, the world needs platforms. Say you want to connect all 200+ countries on Earth with high-speed fiber lines; you can run a cable from each country to every other country (about 21,000 cables, many of them expensively draped across the ocean floor), or you can pick one country (preferably one with both Atlantic and Pacific coasts) and run all your cables there, and then interconnect them.

That's America, the world's global fiber hub. The problem is, America isn't just a platform for fiber interconnections – it's a Great Power that uses its position at the center of the world's fiber networks to surveil and disrupt the world's communications networks.

That's a classic enshittification move on a geopolitical scale. It's not the only one America's made, either.

Consider the US dollar. The dollar is to global commerce what America's fiber head-ends are to the world's data network: a site of essential, (nominally) neutral interchange that is actually a weapon that the US uses to gain advantage over its allies and to punish its enemies:

The world's also got about 200 currencies. For parties in one country to trade with those in another country, the buyer needs to possess a currency the seller can readily spend. The problem is that setting up 21,000 pairwise exchange markets from every currency to every other currency is expensive and cumbersome – traders would have to amass reserves of hundreds of rarely used currencies, or they would have to construct long, brittle, expensive, high-risk chains that convert, say, Thai baht into Icelandic kroner to Brazilian reals and finally into Costa Rican colones.

Thanks to a bunch of complicated maneuvers following World War II, the world settled on the US dollar as its currency platform. Most important international transactions use "dollar clearing" (where goods are priced in USD irrespective of their country of origin) and buyers need only find someone who will convert their currency to dollars in order to buy food, oil, and other essentials.

There are two problems with this system. The first is that America has never treated the dollar as a neutral platform; rather, American leaders have found subtle, deniable ways to use "dollar dominance" to further America's geopolitical agenda, at the expense of other dollar users (you know, "enshittification"). The other problem is that America has become steadily less deniable and subtle in these machinations, finding all kinds of "exceptional circumstances" to use the dollar against dollar users.

America's unabashed dollar weaponization has been getting worse for years, but under Trump, the weaponized dollar has come to constitute an existential risk to the rest of the world, sending them scrambling for alternatives. As November Kelly says, Trump inherited a poker game that was rigged in his favor, but he still flipped over the table because he resents having to pretend to play at all.

Once Trump tried to steal Greenland, it became apparent that the downsides of the dollar far outweigh its upsides. Last month, Christine Lagarde (president of the European Central Bank) made a public announcement on a radio show that Europe "urgently" needed to build its own payment system to avoid the American payment duopoly, Visa/Mastercard.

Arthur T Knackerbracket writes:

The company recently announced a major restructure resulting in job losses, as well as a return-to-office mandate:

Ubisoft's struggles continued this week, with at least 1,200 employees across France and Italy participating in a three-day strike. This comes weeks after the company announced a major restructure resulting in job losses, as well as issued an unpopular return-to-office mandate.

Kicking off on Feb. 10, the industrial action is in response to Ubisoft announcing a massive restructure last month, including the cancellation of six games (once of which was the long-awaited Prince of Persia: The Sands of Time remake). According to the company, the restructure was designed to improve both its growth and its finances.

Ubisoft had already reduced its headcount from almost 21,000 in 2022 to 17,097 by last November, conducting ongoing layoffs intended to reduce costs. Now this new structural shakeup is expected to result in even more job losses. Less than a week after the restructuring was announced, Ubisoft revealed that it would implement a voluntary departure plan at its Paris headquarters to cut approximately 200 employees.

"We were informed of this at the same time as the press — as none of these changes had been discussed during the mandatory consultations with the works councils a few days earlier!" the unions claimed in a joint statement shared on Bluesky.

As such, the unions are calling for an end to "top-down decisions" and cost-cutting plans which they state have employees shouldering the brunt of the consequences. Ubisoft's cost-cutting measures have thus far included the closure of its Halifax and Stockholm studios, as well as job losses at its Abu Dhabi and Massive studios.

"[I]t seems clear today that management has lost sight of the very driving force behind our industry: its workers," wrote the unions. "The announced transformation claims to place games at the heart of its strategy, but without us, these games cannot exist."

In addition to these concerns, unions are also protesting Ubisoft's return to mandatory in-office working revealed with the restructure, which now requires employees to come in five days a week. Demanding an end to "coercive control" of working conditions as well as Ubisoft's "anti-remote-work obsession," the unions have accused the company of creating a work environment unpleasant enough that employees will want to quit.

Original Submission

upstart writes:

Anthropic Launches New Model That Spots Zero Days, Makes Wall Street Traders Lose Their Minds:

Anthropic, the makers of the popular and code-competent chatbot Claude, released a new model Thursday called Claude Opus 4.6. The company is doubling down on coding capabilities, claiming that the new model "plans more carefully, sustains agentic tasks for longer, can operate more reliably in larger codebases, and has better code review and debugging skills to catch its own mistakes."

It seems the model is also pretty good at catching other people's mistakes. According to a report from Axios, Opus 4.6 was able to spot more than 500 previously undisclosed zero-day security vulnerabilities in open-source libraries during its testing period. It also reportedly did so without receiving specific prompting to go hunting for flaws—it just spotted and reported them.

That's a nice change of pace from all of the many developments that have been happening around OpenClaw, an open-source AI agent that most users have been running with Claude Opus 4.5. A number of vibe-coded projects that have come out of the community have had some pretty major security flaws. Maybe Anthropic's upgrade will be able to catch those issues before they become everyone else's problem.

Claude's calling card has been coding for some time now, but it seems Anthropic is looking to make a splash elsewhere with this update. The company said Opus 4.6 will be better at other work tasks like creating PowerPoint presentations and navigating documents in Excel. Seems those features will be key to Cowork, Anthropic's recent project that it is touting as "Claude Code" for non-technical workers.

Arthur T Knackerbracket writes:

Astronomers discover the surprising reason for a star's disappearance

The steady beam of a star twice the size of the sun played a trick on astronomers about a year ago: It vanished.

Then some nine months later, it reappeared in the constellation Monoceros, about 3,200 light-years away in space.

Now researchers think they've solved the mystery of one of the longest star-dimming events ever recorded. The star, called ASASSN-24fw, may have disappeared behind a giant planet with an enormous system of rings, according to new research, blocking most of its light from reaching Earth for nine months.

The steady beam of a star twice the size of the sun played a trick on astronomers about a year ago: It vanished.

Then some nine months later, it reappeared in the constellation Monoceros, about 3,200 light-years away in space.

Now researchers think they've solved the mystery of one of the longest star-dimming events ever recorded. The star, called ASASSN-24fw, may have disappeared behind a giant planet with an enormous system of rings, according to new research, blocking most of its light from reaching Earth for nine months.

[...] The team's top explanation involves a brown dwarf surrounded by humongous rings, similar in shape to Saturn's but vastly larger, eclipsing the star. In this case, the rings are estimated to stretch about 15.8 million miles from the brown dwarf, about half the distance between the sun and Mercury.

As the ring system moved in front of the star, it blocked about 97 percent of ASASSN-24fw's light. By studying changes in the star's brightness and light patterns — methods astronomers use to infer mass and motion — the team estimates the hidden object weighs more than three times as much as Jupiter.

The data also suggest the star itself has leftover material close by, possibly debris from past or ongoing planetary collisions. That is unusual for a star believed to be more than a billion years old.

Journal Reference: Sarang Shah, Jonathan P Marshall, Carlos del Burgo, et al., The nature of ASASSN-24fw's occultation: modelling the event as dimming by optically thick rings around a substellar companion, Monthly Notices of the Royal Astronomical Society, Volume 546, Issue 3, March 2026, staf2251, https://doi.org/10.1093/mnras/staf2251

Original Submission

Today is SN's birthday - we are 12 years old!

The site published its first discussion on the 12 February 2014 but had to be reset a few days later on 15/16 Feb because of software problems which had not been apparent until the community grew. But after 12 years I won't quibble over a few days difference.

This site would not exist without many people who wrote code, configured hardware, tested software, and squashed bugs. It would not be fair to try to name them - I would surely miss many who have been instrumental in getting us where we are today. We initially had a Board comprising of 'shareholders', but today we have a Board of volunteers. The running costs which once were around $6,000 pa are now almost zero due to the generosity of those who donate free hardware and the essential internet connection. Many others over the years have given freely of their time in various roles to keep this site running. No ads, no sponsorship, no commercial pressure.

But the most important people are you - the community. There are still many accounts active that have been with us from the beginning, but those that have joined sometime over the 12 years are equally important and just as welcome. We hope that you all find something of interest in at least some of the stories that we publish. Please keep commenting in them. And if you can, please make the occasional submissions that are essential for our continued operation.

Thank you - this is your site. So I raise my glass to SoylentNews, to this community and, hopefully, to the next 12 years!

upstart writes:

Claude Opus 4.6 spends $20K trying to write a C compiler:

An Anthropic researcher's efforts to get its newly released Opus 4.6 model to build a C compiler left him "excited," "concerned," and "uneasy."

It also left many observers on GitHub skeptical, to say the least.

Nicholas Carlini, a researcher on Anthropic's Safeguards team, detailed the experiment with what he called "agent teams" in a blog that coincided with the official release of Opus 4.6.

He said he "tasked 16 agents with writing a Rust-based C compiler, from scratch, capable of compiling the Linux kernel. After nearly 2,000 Claude Code sessions and $20,000 in API costs, the agent team produced a 100,000-line compiler that can build Linux 6.9 on x86, ARM, and RISC-V."

With agent teams, he said, "multiple Claude instances work in parallel on a shared codebase without active human intervention."

One key task was getting round the need for "an operator to be online and available to work jointly," which we presume means removing the need for Claude Code to wait for a human to tell it what to do next.

"To elicit sustained, autonomous progress, I built a harness that sticks Claude in a simple loop... When it finishes one task, it immediately picks up the next." Imagine if humans took that sort of approach.

Carlini continued: "I leave it up to each Claude agent to decide how to act. In most cases, Claude picks up the 'next most obvious' problem." This threw up a number of lessons, including the need to "write extremely high quality tests."

Readers were also advised to "put yourself in Claude's shoes." That means the "test harness should not print thousands of useless bytes" to make it easier for Claude to find what it needs.

Also, "Claude can't tell time and, left alone, will happily spend hours running tests instead of making progress."

Which might make you feel working with Claude is closer to working with a regular human than you might have thought. But what was the upshot of all of this?

"Over nearly 2,000 Claude Code sessions across two weeks, Opus 4.6 consumed 2 billion input tokens and generated 140 million output tokens, a total cost just under $20,000."

This made it "an extremely expensive project" compared to the priciest Claude Max plans, Carlini said. "But that total is a fraction of what it would cost me to produce this myself – let alone an entire team."

Arthur T Knackerbracket writes:

https://www.zdnet.com/article/personal-digital-sovereignty-choices-free-linux-servers/

You may have noticed that many European Union (EU) governments and agencies, worried about ceding control to untrustworthy US companies, have been embracing digital sovereignty. Those bodies are turning to running their own cloud and services instead of relying on, say, Microsoft 365 or Google Workspace. If you prize your privacy and want to control your own services, you can take that approach as well.

Of course, if you're a techie's techie, you could always run your own cloud. I've been running my own servers for decades. These days, I use AlmaLinux, Rocky Linux, and Ubuntu on my machines.

However, most people don't have many years of Unix/Linux system administration behind them. Fortunately, there are pre-built Linux servers suitable for home and small-business users. With these servers, you still need to be a power user to get the most out of them, but they don't require you to be a Linux expert.

There are three types of ready-to-run Linux server distributions. The first are those that provide software-as-a-service (SaaS) addons and programs. Then there are the distros that focus on providing file server/storage services. Finally, believe it or not, there's one approach meant to replace Windows Server.

1. The privacy-first approach: FreedomBox

FreedomBox, the project initiated by Free Software Foundation (FSF) legal expert Eben Moglen, has matured into Debian's official self-hosting solution.

As Moglen said when he introduced FreedomBox in 2011, "We're building software for smart devices whose engineered purpose is to work together to facilitate free communication among people, safely and securely, beyond the ambition of the strongest power to penetrate. They can make freedom of thought and information a permanent, ineradicable feature of the net that holds our souls."

The platform is now integrated as Debian Linux Blend. This approach enables you to transform a fresh Debian installation into a privacy-focused server via the Plinth web interface.

2. YunoHost: Self-hosting democratized

YunoHost is best described as a "make self‑hosting boring" layer on top of Debian. As its volunteer creators say, "YunoHost is primarily designed for people who want things to 'just work.'"

Similar to Freedom Box, YunoHost functions as both a standalone operating system and a package you can install on an existing Debian installation. Unlike FreedomBox, which can be scaled up for a small business, the YunoHost crew warns, "YunoHost is not designed to 'scale' in the traditional sense. It is intended for a relatively modest number of user accounts and simultaneous users." So, a few dozen users? No problem. A few hundred? No, just no.

YunoHost comes with a small, integrated server stack. Everything else is added from its catalog. On a fresh YunoHost install, you get these main components by default: a web admin interface and a user portal for installing and logging in to all the applications. This setup is supported by Nginx as the web server and reverse proxy, with SSOwat for single sign-on to all installed web apps.

You can also install an email server stack from the start. Your default programs are Postfix for the Simple Mail Transfer Protocol (SMTP) server, Dovecot as the Internet Message Access Protocol (IMAP) server, and Rspamd, with DomainKeys Identified Mail (DKIM) handling for spam filtering and mail authentication. As e-mail server programs go, these are the easiest to manage, and YunoHost does a great job of installing them.

However, speaking as someone who's been running email servers for decades, setting them up and managing them on the internet is hard work. You'll need to set up a proper domain, DNS records (MX, SPF, DKIM, DMARC) with a static IP address. If your eyes just glazed over, don't try running your own email server.

Like FreedomBox, YunoHost is completely free.

Arthur T Knackerbracket writes:

https://www.techspot.com/news/111247-hiroshima-scientists-crack-code-3d-printing-tungsten-carbide.html

In a breakthrough that could reshape how tools for harsh environments are made, scientists at Hiroshima University have developed a method to 3D print one of the toughest materials used in industry: tungsten carbide – cobalt. The advance overcomes a long-standing challenge in additive manufacturing – how to shape ultra-hard composites without damaging their internal structure.

The university's team reports that their approach centers on controlled "softening" of the material rather than complete melting. The process, known as hot-wire laser irradiation, reshapes tungsten carbide while maintaining its exceptional hardness and minimizing defects – an achievement that could transform how cutting, drilling, and construction tools are manufactured.

Unlike most 3D printing workflows, which rely on fully melting metal powders or rods, the Hiroshima group used a laser to heat tungsten carbide rods just enough to make them pliable. This prevented grain growth and decomposition that often occur at full melting temperatures.

To bond multiple printed layers securely, researchers added a nickel-based alloy as an intermediate layer within the build. The result: dense parts with a measured surface hardness exceeding 1,400 HV, approaching the hardness of gemstones like sapphire.

Assistant Professor Keita Marumoto of Hiroshima University's Graduate School of Advanced Science and Engineering described the technique as an entirely new approach to forming metallic materials. He noted that, while the current work focused on cemented carbides such as WC – Co, the same principle could potentially apply to other difficult-to-manufacture compounds.

Traditional approaches involve sintering powdered materials in molds, which limits geometrical complexity and generates substantial waste. Additive manufacturing could, in theory, solve both problems – if the material could survive the process.

While the achievement represents a leap forward, the research group acknowledges that their work remains ongoing. They are fine-tuning the process to eliminate occasional cracking and plan to test how far the technique can be extended to more intricate geometries.

If successful, additive manufacturing could soon produce complex industrial tools that combine durability with material efficiency – an outcome long out of reach for engineers working with ultra-hard composites.

Original Submission

fliptop writes:

Ford Motor Company on Feb. 10 reported fourth-quarter revenue 2025 of $45.9 billion, a 5 percent year-over-year decline that led to its largest earnings miss since the same quarter in 2021:

Ford posted a net loss of $11.1 billion in the quarter and earnings per share of $0.13, well below analysts' forecast of $0.18. In the year-ago quarter, Ford posted net income of $1.8 billion and earnings per share of $0.45. The Dearborn, Michigan-based automaker's full-year revenue of $187.3 billion was up from $185 billion in 2024, marking the fifth straight year of revenue growth despite the challenging fourth quarter.

Its net loss for the year, however, was $8.2 billion, versus net income of $5.9 billion in 2024.

Ford CEO Jim Farley said during a conference call with analysts that the impact from a fire at the Novelis aluminum plant in Oswego, New York—a major aluminum supplier for the automaker's F-series pickup trucks—and unexpected changes to tariff credits for auto parts resulted in costs of roughly $4 billion.

[...] Ford also provided full-year guidance for 2026 of adjusted earnings before interest and taxes of $8–10 billion, up from the $6.8 billion reported in 2025, and in line with the FactSet analyst estimate of $8.78 billion.

From Road & Track:

Ford is not alone in its decision to take a step back from its lofty plans for electric vehicles, as the entire auto industry grapples with slowing demand for battery-powered cars and trucks, but a recent financial report from the Dearborn-based automaker spells out just how painful the situation has been for the company's bank accounts.

Related:

• Ford EV Battery Joint Venture Plant Lays Off All Workers
• Ford Cancels Electric F-150
• F-150 Lightning Production Halted Indefinitely As Ford Bets On Gas Trucks Again

Original Submission

hubie writes:

Four baby planets show how super-Earths and sub-Neptunes form:

Thanks to the discovery of thousands of exoplanets to date, we know that planets bigger than Earth but smaller than Neptune orbit most stars. Oddly, our sun lacks such a planet. That's been a source of frustration for planetary scientists, who can't study them in as much detail as they'd like, leaving one big question: How did these planets form?

Now we know the answer.

An international team of astrophysicists from UCLA and elsewhere has witnessed four baby planets in the V1298 Tau system in the process of becoming super-Earths and sub-Neptunes. The findings are published in the journal Nature.

"I'm reminded of the famous 'Lucy' fossil, one of our hominid ancestors that lived 3 million years ago and was one of the 'missing links' between apes and humans," said UCLA professor of physics and astronomy and second author Erik Petigura. "V1298 Tau is a critical link between the star- and planet-forming nebulae we see all over the sky, and the mature planetary systems that we have now discovered by the thousands."

Planets form when a cloud of gas and dust, called a nebula, contracts under the force of gravity into a young star and a swirling disk of matter called a protoplanetary disk. Planets form from this disk of gas, but it's a messy process. There are many ways a planet can grow or shrink in size during its infancy --- a period of a few hundred million years. This led to major questions about why so many mature planets were between the sizes of Earth and Neptune.

The star V1298 Tau is only about 20 million years old compared to our 4.5-billion-year-old sun. Expressed in human terms, it's equivalent to a 5-month-old baby. Four giant, rapidly evolving planets between the sizes of Neptune and Jupiter orbit the star, but unlike growing babies, the new research shows that these planets are contracting in size and are steadily losing their atmospheres. Petigura and co-author Trevor David at the Flatiron Institute led the team that first discovered the planets in 2019.

"What's so exciting is that we're seeing a preview of what will become a very normal planetary system," said John Livingston, the study's lead author from the Astrobiology Center in Tokyo, Japan. "The four planets we studied will likely contract into 'super-Earths' and 'sub-Neptunes'—the most common types of planets in our galaxy, but we've never had such a clear picture of them in their formative years."

[...] Once they sorted out the shapes and timing of the orbits of the four planets, the researchers could make sense of how the planets tugged on each other due to gravity, sometimes slowing down and sometimes speeding up, and leading to transits, sometimes occurring early and other times late. These transit and timing variations allowed the team to measure the masses of all four planets for the first time, which is akin to weighing them.

The shocking result? Despite being 5 to 10 times the radius of Earth, the planets had masses only 5 to 15 times larger than Earth. This means they are very low-density, comparable to Styrofoam, whereas the Earth has the density of rock.

"The unusually large radii of young planets led to the hypothesis that they have very low densities, but this had never been measured," said Trevor David, a co-author from the Flatiron Institute who led the initial discovery of the system in 2019. "By weighing these planets for the first time, we have provided the first observational proof. They are indeed exceptionally 'puffy,' which gives us a crucial, long-awaited benchmark for theories of planet evolution."

"Our measurements reveal they are incredibly lightweight — some of the least dense planets ever found. It's a critical step that turns a long-standing theory about how planets mature into an observed reality," said Livingston.

[...] "These planets have already undergone a dramatic transformation, rapidly losing much of their original atmospheres and cooled faster than what we'd expect from standard models," said James Owen, a co-author from Imperial College London who led the theoretical modeling. "But they're still evolving. Over the next few billion years, they will continue to lose their atmosphere and shrink significantly, transforming into the compact systems of super-Earths and sub-Neptunes we see throughout the galaxy."

Journal Reference: Livingston, J.H., Petigura, E.A., David, T.J. et al. A young progenitor for the most common planetary systems in the Galaxy. Nature 649, 310–314 (2026). https://doi.org/10.1038/s41586-025-09840-z

Original Submission