SoylentNews

upstart writes:

NASA boss says US should have 'village' on Moon in a decade:

IAC 2025 If the USA's space strategy succeeds, it will run a "village" on the moon in a decade, NASA administrator Sean Duffy told the International Aeronautical Congress (IAC) in Sydney today.

Duffy appeared in a session featuring the heads of space agencies from the USA, China, Japan, India, Europe, and Canada. Readers will likely have noted the absence of Russia, a longtime space player, from that list.

The NASA boss seemingly hinted at one reason Russia's space boss is not at the Congress when he said the USA "comes in peace" to space. "We have not been in the business of taking people's land," Duffy said.

Asked what success looks like for NASA in a decade, Duffy said "We are going to have sustained human life on the moon. Not just an outpost, but a village." And a nuclear-powered village at that after NASA recently issued an RFI seeking commercial help to build a nuclear reactor on Luna

Duffy also predicted that a decade from now NASA will also have "made leaps and bounds on our mission to get to Mars" and "be on the cusp of putting human boots on Mars."

The theme for this year's IAC is "Sustainable Space: Resilient Earth". Duffy's take on that is how to sustain human life in space, an objective he said is NASA's prerogative because it alone among US government agencies has a remit for exploration. He pointed out that other US government agencies have the job of considering terrestrial stability and earthly resilience, and that NASA must focus on exploration,

The other space agency heads at the event took a different view.

When European Space Agency (ESA) director general Josef Ashbacher had his turn on stage, he offered a very different vision of sustainability by pointing out that the agency he leads freely shares data from its earth observation satellites. "I am glad that we at ESA are working for the betterment of the planet," he said.

V Narayanan, the chair of India's Space Research Organization (ISRO) said ensuring food and water resources is his agency's top goal in space. Lisa Campbell, the President of the Canadian Space Agency said that when her country first orbited an earth observation satellite it had to pay private sector organizations to use its data. "Now they want it," she said, before announcing CA$5 million ($3.6 million) to fund studies on biodiversity from space. The president of the Japan Aerospace Exploration Agency (JAXA), Dr Hiroshi Yamakawa, reminded attendees that Japan recently launched its third greenhouse gas observation satellite.

Anonymous Coward writes:

https://www.bleepingcomputer.com/news/security/cisa-warns-of-critical-linux-sudo-flaw-exploited-in-attacks/

Hackers are actively exploiting a critical vulnerability (CVE-2025-32463) in the sudo package that enables the execution of commands with root-level privileges on Linux operating systems.

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added this vulnerability to its Known Exploited Vulnerabilities (KEV) catalog, describing it as "an inclusion of functionality from untrusted control sphere."

CISA has given federal agencies until October 20 to apply the official mitigations or discontinue the use of sudo.

A local attacker can exploit this flaw to escalate privileges by using the -R (--chroot) option, even if they are not included in the sudoers list, a configuration file that specifies which users or groups are authorized to execute commands with elevated permissions.

Sudo ("superuser do") allows system administrators to delegate their authority to certain unprivileged users while logging the executed commands and their arguments.

Officially disclosed on June 30, CVE-2025-32463 affects sudo versions 1.9.14 through 1.9.17 and has received a critical severity score of 9.3 out of 10.

"An attacker can leverage sudo's -R (--chroot) option to run arbitrary commands as root, even if they are not listed in the sudoers file," explains the security advisory.

Rich Mirch, a researcher at cybersecurity services company Stratascale who discovered CVE-2025-32463, noted that the issue impacts the default sudo configuration and can be exploited without any predefined rules for the user.

On July 4, Mirch released a proof-of-concept exploit for the CVE-2025-32463 flaw, which has existed since June 2023 with the release of version 1.9.14.

However, additional exploits have circulated publicly since July 1, likely derived from the technical write-up.

CISA has warned that the CVE-2025-32463 vulnerability in sudo is being exploited in real-world attacks, although the agency has not specified the types of incidents in which it has been leveraged.

Organizations worldwide are advised to use CISA's Known Exploited Vulnerabilities catalog as a reference for prioritizing patching and implementing other security mitigations.

Original Submission

janrinok writes:

https://phys.org/news/2025-09-photodiode-germanium-key-chip.html

Programmable photonics devices, which use light to perform complex computations, are emerging as a key area in integrated photonics research. Unlike conventional electronics that transmit signals with electrons, these systems use photons, offering faster processing speeds, higher bandwidths, and greater energy efficiency. These advantages make programmable photonics well-suited for demanding tasks like real-time deep learning and data-intensive computing.

A major challenge, however, lies in the use of power monitors. These sensors must constantly track the optical signal's strength and provide the necessary feedback for tuning the chip's components as required. However, existing on-chip photodetectors designed for this purpose face a fundamental tradeoff. They either have to absorb a significant amount of the optical signal to achieve a strong reading, which degrades the signal's quality, or they lack the sensitivity to operate at the low power levels required without needing additional amplifiers.

As reported in Advanced Photonics, Yue Niu and Andrew W. Poon from The Hong Kong University of Science and Technology have addressed this challenge by developing a germanium-implanted silicon waveguide photodiode. Their approach overcomes the tradeoffs that have hindered existing on-chip power monitoring technologies.

A waveguide photodiode is a small light detector that can be integrated directly into an optical waveguide, which confines and transports light. Its purpose is to convert a small portion of the light traveling through the waveguide into an electrical signal that can be measured via more conventional electronics. One way to enhance this conversion is through ion implantation, a process that introduces controlled defects into the photodiode's silicon structure by bombarding it with ions.

If executed properly, these defects can absorb photons with energies too low for pure silicon, enabling the photodiode to detect light across a broader range of wavelengths.

Previous attempts to build such detectors used boron, phosphorus, or argon ions. While these approaches improved performance in some respects, they also introduced many free carriers into the silicon lattice, which in turn degraded optical performance. In contrast, the team implanted germanium ions. Germanium, a Group IV element like silicon, can replace silicon atoms in the crystal structure without introducing significant numbers of free carriers. This substitution allows the device to extend its sensitivity without compromising signal quality.

AnonTechie writes:

404MEDIA

ICE to Buy Tool that Tracks Locations of Hundreds of Millions of Phones Every Day

Documents show that ICE has gone back on its decision to not use location data remotely harvested from peoples' phones. The database is updated every day with billions of pieces of location data.

Immigration and Customs Enforcement (ICE) has bought access to a surveillance tool that is updated every day with billions of pieces of location data from hundreds of millions of mobile phones, according to ICE documents reviewed by 404 Media.

The documents explicitly show that ICE is choosing this product over others offered by the contractor's competitors because it gives ICE essentially an "all-in-one" tool for searching both masses of location data and information taken from social media. The documents also show that ICE is planning to once again use location data remotely harvested from peoples' smartphones after previously saying it had stopped the practice.

Surveillance contractors around the world create massive datasets of phones', and by extension people's movements, and then sell access to the data to government agencies. In turn, U.S. agencies have used these tools without a warrant or court order.

"The Biden Administration shut down DHS's location data purchases after an inspector general found that DHS had broken the law. Every American should be concerned that [the current administration's] hand-picked security force is once again buying and using location data without a warrant," Senator Ron Wyden told 404 Media in a statement.

Original Submission

AnonTechie writes:

The AI industry has made major promises about its tech boosting the productivity of developers, allowing them to generate copious amounts of code with simple text prompts.

But those claims appear to be massively overblown, as The Register reports, with researchers finding that productivity gains are modest at best — and at worst, that AI can actually slow down human developers.

In a new report, management consultants Bain & Company found that despite being "one of the first areas to deploy generative AI," the "savings have been unremarkable" in programming.

"Generative AI arrived on the scene with sky-high expectations, and many companies rushed into pilot projects," the report reads. "Yet the results haven't lived up to the hype."

First off, "developer adoption is low" even among the companies that rolled out AI tools, the management consultancy found.

Worse yet, while some assistants saw "ten to 15 percent productivity boosts," the savings most of the time "don't translate into positive returns."

It's yet another warning shot, highlighting concerns that even in one of the most promising areas, the AI industry is struggling to live up to its enormous hype. That's despite companies pouring untold billions of dollars into its development, with analysts openly fretting about an enormous AI bubble getting closer to popping.

futurism.com

Original Submission

janrinok writes:

https://phys.org/news/2025-09-insect-cultivated-proteins-healthier-greener.html

Reducing industrial animal use can help to shrink our carbon footprint and boost health—but doing so means we need nutritious meat alternatives that are also tasty and affordable.

The researchers say that by using combinations of different proteins from plants, fungi, insects, microbial fermentation, and cultivated meat, we could create tasty, nutritious, and sustainable alternatives to animal products.

As well as tackling environmental concerns, hybrids could also help to address the health and ethical impact of livestock farming such as animal welfare, zoonotic disease, and antimicrobial resistance.

"Hybrid foods could give us a delicious taste and texture without breaking the bank or the planet," said first author Prof David L. Kaplan from Tufts University in the US. "Using protein alternatives needn't necessarily come with financial, taste, or nutritional costs."

For example, by drawing on the fibrous texture of mycelium, the sensory and nutritional qualities of cultivated meat, the nutrition and sustainability of insects, the proteins, pigments, enzymes, and flavors from microbial fermentation, and the abundance and low cost of plants, hybrids could combine the best of each protein source, say the authors.

But to make this happen, the researchers call for regulatory review and academic and industry cooperation to overcome hurdles and find the best possible protein combinations for our health, sensory, environmental, and cost needs.

"To succeed, we need research and cooperation across science, industry, and regulators to improve quality, scale production, and earn consumer trust," added Prof Kaplan.

The researchers investigated different protein sources: plants (for example, soy products like tofu), insects (processed into flours and blended into foods), mycelium-based products (such as vegan commercial meat analogs), cultivated meat grown in bioreactors, and microbial fermentation products (such as proteins, pigments, enzymes, and flavors).

They assessed the strengths and weaknesses of each protein source and considered how to harness the best qualities of each—both with and without animal meat. For example, while plant proteins are cheap and scalable, they often lack the flavor and texture of meat. Meanwhile, cultivated meat more closely mimics animal meat but is expensive and hard to scale. Mycelium can add natural texture, while insects offer high nutrition with a low environmental footprint.

The researchers reviewed various combinations to compare their sensory and nutritional profiles, consumer acceptance, affordability, and scalability.

They found that while every protein source has drawbacks, combining them can overcome many of these limitations. In the short term, plant–mycelium hybrids appear most economically viable because they are scalable, nutritious, and already used in commercial products.

upstart writes:

Complex knots can actually be easier to untie than simple ones:

Why is untangling two small knots more difficult than unravelling one big one? Surprisingly, mathematicians have found that larger and seemingly more complex knots created by joining two simpler ones together can sometimes be easier to undo, invalidating a conjecture posed almost 90 years ago.

"We were looking for a counterexample without really having an expectation of finding one, because this conjecture had been around so long," says Mark Brittenham at the University of Nebraska at Lincoln. "In the back of our heads, we were thinking that the conjecture was likely to be true. It was very unexpected and very surprising. "

Mathematicians like Brittenham study knots by treating them as tangled loops with joined ends. One of the most important concepts in knot theory is that each knot has an unknotting number, which is the number of times you would have to sever the string, move another piece of the loop through the gap and then re-join the ends before you reached a circle with no crossings at all – known as the "unknot".

Calculating unknotting numbers can be a very computationally intensive task, and there are still knots with as few as 10 crossings that have no solution. Because of this, it can be helpful to break knots down into two or more simpler knots to analyse them, with those that can't be split any further known as prime knots, analogous to prime numbers.

But a long-standing mystery is whether the unknotting numbers of the two knots added together would give you the unknotting number of the larger knot. Intuitively, it might make sense that a combined knot would be at least as hard to undo as the sum of its constituent parts, and in 1937, it was conjectured that undoing the combined knot could never be easier.

Now, Brittenham and Susan Hermiller, also at the University of Nebraska at Lincoln, have shown that there are cases when this isn't true. "The conjecture's been around for 88 years and as people continue not to find anything wrong with it, people get more hopeful that it's true," says Hermiller. "First, we found one, and then quickly we found infinitely many pairs of knots for whom the connected sum had unknotting numbers that were strictly less than the sum of the unknotting numbers of the two pieces."

"We've shown that we don't understand unknotting numbers nearly as well as we thought we did," says Brittenham. "There could be – even for knots that aren't connected sums – more efficient ways than we ever imagined for unknotting them. Our hope is that this has really opened up a new door for researchers to start exploring."

While finding and checking the counterexamples involved a combination of existing knowledge, intuition and computing power, the final stage of checking the proof was done in a decidedly more simple and practical manner: tying the knot with a piece of rope and physically untangling it to show that the researchers' predicted unknotting number was correct.

Andras Juhasz at the University of Oxford, who previously worked with AI company DeepMind to prove a different conjecture in knot theory, says that he and the company had tried unsuccessfully to crack this latest problem about additive sets in the same way, but with no luck.

"We spent at least a year or two trying to find a counterexample and without success, so we gave up," says Juhasz. "It is possible that for finding counterexamples that are like a needle in a haystack, AI is maybe not the best tool. This was a hard-to-find counterexample, I believe, because we searched pretty hard."

Despite there being many practical applications for knot theory, from cryptography to molecular biology, Nicholas Jackson at the University of Warwick, UK, is hesitant to suggest that this new result can be put to good use. "I guess we now understand a little bit more about how circles work in three dimensions than we did before," he says. "A thing that we didn't understand quite so well a couple of months ago is now understood slightly better."

Reference:

Journal Reference:
Brittenham, Mark, Hermiller, Susan. Unknotting number is not additive under connected sum, (DOI: 10.48550/arXiv.2506.24088)

See also:

• Unknotting number is not additive under connected sum
• A century-old knot conjecture was wrong all along [24:46 --JE]

Original Submission

upstart writes:

Huawei's Ternary Logic Breakthrough: A Game-Changer or Just Hype?:

[Editor's Comment: The source reads as though it could have been created by AI, nevertheless it is an interesting topic and worth a discussion.--JR]

Huawei's recent patent for 'ternary logic' represents a potential breakthrough in chip technology by utilizing a three-state logic system consisting of -1, 0, and 1, rather than the traditional binary system of 0 and 1. This innovative approach could substantially reduce the number of transistors required on a chip, leading to lower energy consumption, particularly in power-hungry AI applications. The ternary logic patent, filed in September 2023 and recently disclosed, may offer significant advantages in processing efficiency and hardware design, addressing some of the physical limits faced by current chip technologies.

Ternary computing is not a novel concept; the first ternary computer was developed in 1958 at Moscow State University, indicating the feasibility of utilizing more than two states in computational logic (source). However, binary logic became the industry standard due to its simplicity and the development of compatible technologies. Huawei's pursuit of ternary logic comes amidst US sanctions, which have pressured the company to explore alternative technological paths. By reducing reliance on traditional chip designs, Huawei aims to innovate in a constrained environment and potentially gain a competitive edge in the AI and semiconductor sectors.

The commercial viability of Huawei's ternary logic chip presents an intriguing yet complex scenario for the tech industry. Ternary logic, which utilizes three states instead of the usual binary system's two, promises significant advancements in chip technology. By potentially reducing the number of transistors required, it could lead to decreases in both manufacturing costs and energy consumption, particularly in power-hungry AI applications. However, the road to commercial viability is laden with challenges. The tech industry must grapple with the transition from a binary-dominated ecosystem to one that could incorporate ternary systems. This includes revamping software and programming approaches to leverage the benefits of the new computing structure. Furthermore, the ability to mass-produce these chips economically and reliably remains unproven, leaving questions about whether the technology can achieve a cost-effective scale .

If successful, Huawei's ternary logic patent could disrupt current computing paradigms and lead to reduced energy consumption in AI technology, aligning with broader trends towards sustainability. The potential of such technology to alter chip design and improve AI efficiency could have far-reaching implications not only for Huawei but for the tech industry at large. Moreover, by possibly circumventing some effects of international sanctions, Huawei's efforts symbolize a form of technological resilience and ingenuity amid geopolitical challenges

Ternary logic represents a novel approach in computing, differentiating itself from the conventional binary logic by utilizing three distinct values: -1, 0, and 1. This trinary system aims to encode information more efficiently, offering potential reductions in the complexity and power usage of AI chips. Unlike the binary system that utilizes two states (0 and 1) to process tasks, ternary logic could lead to less energy-intensive data processing, ultimately decreasing the overall power consumption of AI-driven technologies. This innovation heralds a shift from traditional methodologies, potentially streamlining both hardware requirements and computational resource demands.

The implementation of ternary logic in modern computing could signify a breakthrough in addressing the issues associated with current chip designs. With chip manufacturing approaching its physical limits, the trinary system provides an alternative pathway to enhance processing capabilities without exponentially increasing transistor counts. Huawei's recent patent reflects this innovative direction, aiming to solve power consumption dilemmas while navigating international sanctions and fostering technological advancements in AI. Embedded in this development is Huawei's strategic response to geopolitical challenges, exemplified by their proactive patent applications that emphasize reducing dependence on traditional binary constraints.

Original Submission

upstart writes:

Experts Alarmed That AI Is Now Producing Functional Viruses:

In real world experiments, a team of Stanford researchers demonstrated that a virus with AI-written DNA could target and kill specific bacteria, they announced in a study last week. It opened up a world of possibilities where artificial viruses could be used to cure diseases and fight infections.

But experts say it also opened a Pandora's box. Bad actors could just as easily use AI to crank out novel bioweapons, keeping doctors and governments on the backfoot with the outrageous pace at which these viruses can be designed, warn Tal Feldman, a Yale Law School student who formerly built AI models for the federal government, and Jonathan Feldman, a computer science and biology researcher at Georgia Tech (no word on whether the two are related).

"There is no sugarcoating the risks," the pair warned in a piecefor the Washington Post. "We're nowhere near ready for a world in which artificial intelligence can create a working virus, but we need to be — because that's the world we're now living in."

In the study, the Stanford researchers used an AI model called Evo to invent DNA for a bacteriophage, a virus that infects bacteria. Unlike a general purpose large language model like ChatGPT, which is trained on written language, Evo was exclusively trained on millions of bacteriophage genomes.

They focused on an extensively studied phage called phiX174, which is known to infect strains of the bacteria E. coli. Using the EVO AI model, the team came up with 302 candidate genomes based on phiX174 and put them to the test by using the designs to chemically assemble new viruses.

Sixteen of them worked, infecting and killing the E. coli strains. Some of them were even deadlier than the natural form of the virus.

But "while the Stanford team played it safe, what's to stop others from using open data on human pathogens to build their own models?" the two Feldmans warned. "If AI collapses the timeline for designing biological weapons, the United States will have to reduce the timeline for responding to them. We can't stop novel AI-generated threats. The real challenge is to outpace them."

That means using the same AI tech to design antibodies, antivirals, and vaccines. This work is already being done to some extent, but the vast amounts of data needed to accelerate such pioneering research "is siloed in private labs, locked up in proprietary datasets or missing entirely."

"The federal government should make building these high-quality datasets a priority," the duo opined.

From there, the federal government would need to build the necessary infrastructure to manufacture these AI-designed medicines, since the "private sector cannot justify the expense of building that capacity for emergencies that may never arrive," they argue.

Finally, the Food and Drug Administration's sluggish and creaking regulatory framework would need an overhaul. (Perhaps in a monkey's paw of such an overhaul, the FDA said it's using AI to speed-run the approval of medications.)

"Needed are new fast-tracking authorities that allow provisional deployment of AI-generated countermeasures and clinical trials, coupled with rigorous monitoring and safety measures," they said.

The serious risks posed by AI virus generation shouldn't be taken lightly. Yet, it's worth noting that the study in question hasn't made it out of peer review yet and we still don't have a full picture of how readily someone could replicate the work the scientists did.

But with agencies like the Centers for Disease Control and Prevention being gutted, and vaccines and other medical interventions being attacked by a health-crank riddled administration, there's no denying that the country's medical policy and infrastructure is in a bad place. That said, when you consider that the administration is finding any excuse to rapidly deploy AI in every corner of the government, it's worth treading lightly when we ask for more.

More on synthetic biology:Scientists Debate Whether to Halt Type of Research That Could Destroy All Life on Earth

hubie writes:

Triple-fermented Belgian beers have the longest-lasting foam; single-fermented lagers have the shortest:

For many beer lovers, a nice thick head of foam is one of life's pure pleasures, and the longer that foam lasts, the better the beer-drinking experience. A team of Swiss researchers spent seven years studying why some beer foams last longer than others and found that the degree of fermentation—i.e., whether a given beer has been singly, doubly, or triply fermented—is crucial, according to a new paper published in the journal Physics of Fluids.

[...] Individual bubbles typically form a sphere because that's the shape with the minimum surface area for any volume and hence is the most energy-efficient. One reason for the minimizing principle when it comes to a bubble's shape is that many bubbles can then tightly pack together to form a foam. But bubbles "coarsen" over time, the result of gravity pulling down on the liquid and thinning out the walls. Eventually, they start to look more like soccer balls (polyhedrons). In a coarsening foam, smaller bubbles are gradually absorbed by larger ones. There is less and less liquid to separate the individual bubbles, so they press together to fill the space.

This "jamming" is why foams are typically far more rigid than their gas (95 percent) and liquid (5 percent) components. The more tightly the bubbles jam together, the less they can move around and the greater the pressure inside them becomes, giving them properties of a solid.

Various factors can affect foam stability. For instance, in 2019, Japanese researchers investigated a phenomenon known as "collective bubble collapse," or CBC, in which breaking one bubble at the edge of a foam results in a cascading effect as the breakage spreads to other bubbles in the foam. They identified two distinct mechanisms for the resulting CBCs: a so-called "propagating mode," in which a broken bubble is absorbed into the liquid film, and a "penetrating mode," in which the breakage of a bubble causes droplets to shoot off and hit other bubbles, causing them to break in turn.

Higher levels of liquid in the foam slowed the spread of the collapse, and changing the viscosity of the fluid had no significant impact on how many bubbles broke in the CBC. Many industrial strategies for stabilizing foams rely on altering the viscosity; this shows those methods are ineffective. The researchers suggest focusing instead on using several different surfactants in the mixture. This would strengthen the resulting film to make it more resistant to breakage when hit by flying droplets.

However, as the authors of this latest paper note, "Most beers are not detergent solutions (though arguably some may taste that way)." They were inspired by a Belgian brewer's answer when they asked how he controlled fermentation: "By watching the foam." A stable foam is considered to be a sign of successful fermentation. The authors decided to investigate precisely how the various factors governing foam stability might be influenced by the fermentation process.

hubie writes:

How the von Neumann bottleneck is impeding AI computing:

Most computers are based on the von Neumann architecture, which separates compute and memory. This arrangement has been perfect for conventional computing, but it creates a data traffic jam in AI computing.

AI computing has a reputation for consuming epic quantities of energy. This is partly because of the sheer volume of data being handled. Training often requires billions or trillions of pieces of information to create a model with billions of parameters. But that's not the whole reason — it also comes down to how most computer chips are built.

Modern computer processors are quite efficient at performing the discrete computations they're usually tasked with. Though their efficiency nosedives when they must wait for data to move back and forth between memory and compute, they're designed to quickly switch over to work on some unrelated task. But for AI computing, almost all the tasks are interrelated, so there often isn't much other work that can be done when the processor gets stuck waiting, said IBM Research scientist Geoffrey Burr.

In that scenario, processors hit what is called the von Neumann bottleneck, the lag that happens when data moves slower than computation. It's the result of von Neumann architecture, found in almost every processor over the last six decades, wherein a processor's memory and computing units are separate, connected by a bus. This setup has advantages, including flexibility, adaptability to varying workloads, and the ability to easily scale systems and upgrade components. That makes this architecture great for conventional computing, and it won't be going away any time soon.

But for AI computing, whose operations are simple, numerous, and highly predictable, a conventional processor ends up working below its full capacity while it waits for model weights to be shuttled back and forth from memory. Scientists and engineers at IBM Research are working on new processors, like the AIU family, which use various strategies to break down the von Neumann bottleneck and supercharge AI computing.

The von Neumann bottleneck is named for mathematician and physicist John von Neumann, who first circulated a draft of his idea for a stored-program computer in 1945. In that paper, he described a computer with a processing unit, a control unit, memory that stored data and instructions, external storage, and input/output mechanisms. His description didn't name any specific hardware — likely to avoid security clearance issues with the US Army, for whom he was consulting. Almost no scientific discovery is made by one individual, though, and von Neumann architecture is no exception. Von Neumann's work was based on the work of J. Presper Eckert and John Mauchly, who invented the Electronic Numerical Integrator and Computer (ENIAC), the world's first digital computer. In the time since that paper was written, von Neumann architecture has become the norm.

"The von Neumann architecture is quite flexible, that's the main benefit," said IBM Research scientist Manuel Le Gallo-Bourdeau. "That's why it was first adopted, and that's why it's still the prominent architecture today."

upstart writes:

This black hole flipped its magnetic field:

The magnetic field swirling around an enormous black hole, located about 55 million light-years from Earth, has unexpectedly switched directions. This dramatic reversal challenges theories of black hole physics and provides scientists with new clues about the dynamic nature of these shadowy giants.

The supermassive black hole, nestled in the heart of the M87 galaxy, was first imaged in 2017. Those images revealed, for the first time, a glowing ring of plasma­ — an accretion disk — encircling the black hole, dubbed M87*. At the time, the disk's properties, including those of the magnetic field embedded in the plasma, matched theoretical predictions.

But observations of the accretion disk in the years that followed show that its magnetic field is not as stable as it first seemed, researchers report in a paper to appear in Astronomy & Astrophysics. In 2018, the magnetic field shifted and nearly disappeared. By 2021, the field had completely flipped direction.

"No theoretical models we have today can explain this switch," says study coauthor Chi-kwan Chan, an astronomer at Steward Observatory in Tucson. The magnetic field configuration, he says, was expected to be stable due to the black hole's large mass — roughly 6 billion times as massive as the sun, making it over a thousand times as hefty as the supermassive black hole at the center of the Milky Way.

In the new study, astronomers analyzed images of the accretion disk around M87* compiled by the Event Horizon Telescope, a global network of radio telescopes. The scientists focused on a specific component that's sensitive to magnetic field orientation called polarized light, which consists of light waves all oscillating in a particular direction.

By comparing the polarization patterns over the years, the astronomers saw that the magnetic field reversed direction. Magnetic fields around black holes are thought to funnel in material from their surrounding disks. With the new findings, astronomers will have to rethink their understanding of this process.

While researchers don't yet know what caused the flip in this disk's magnetic field, they think it could have been a combination of dynamics within the black hole and external influences.

"I was very surprised to see evidence for such a significant change in M87's magnetic field over a few years," says astrophysicist Jess McIver of the University of British Columbia in Vancouver, who was not involved with the research. "This changes my thinking about the stability of supermassive black holes and their environments."

Original Submission

hubie writes:

Expert calls security advice "unfairly outsourcing the problem to Anthropic's users"

On Tuesday [September 9, 2025], Anthropic launched a new file creation feature for its Claude AI assistant that enables users to generate Excel spreadsheets, PowerPoint presentations, and other documents directly within conversations on the web interface and in the Claude desktop app. While the feature may be handy for Claude users, the company's support documentation also warns that it "may put your data at risk" and details how the AI assistant can be manipulated to transmit user data to external servers.

The feature, awkwardly named "Upgraded file creation and analysis," is basically Anthropic's version of ChatGPT's Code Interpreter and an upgraded version of Anthropic's "analysis" tool. It's currently available as a preview for Max, Team, and Enterprise plan users, with Pro users scheduled to receive access "in the coming weeks," according to the announcement.

The security issue comes from the fact that the new feature gives Claude access to a sandbox computing environment, which enables it to download packages and run code to create files. "This feature gives Claude Internet access to create and analyze files, which may put your data at risk," Anthropic writes in its blog announcement. "Monitor chats closely when using this feature."

According to Anthropic's documentation, "a bad actor" manipulating this feature could potentially "inconspicuously add instructions via external files or websites" that manipulate Claude into "reading sensitive data from a claude.ai connected knowledge source" and "using the sandbox environment to make an external network request to leak the data."

This describes a prompt injection attack, where hidden instructions embedded in seemingly innocent content can manipulate the AI model's behavior—a vulnerability that security researchers first documented in 2022. These attacks represent a pernicious, unsolved security flaw of AI language models, since both data and instructions in how to process it are fed through as part of the "context window" to the model in the same format, making it difficult for the AI to distinguish between legitimate instructions and malicious commands hidden in user-provided content.

[...] Anthropic is not completely ignoring the problem, however. The company has implemented several security measures for the file creation feature. For Pro and Max users, Anthropic disabled public sharing of conversations that use the file creation feature. For Enterprise users, the company implemented sandbox isolation so that environments are never shared between users. The company also limited task duration and container runtime "to avoid loops of malicious activity."

[...] Anthropic's documentation states the company has "a continuous process for ongoing security testing and red-teaming of this feature." The company encourages organizations to "evaluate these protections against their specific security requirements when deciding whether to enable this feature."

[...] That kind of "ship first, secure it later" philosophy has caused frustrations among some AI experts like Willison, who has extensively documented prompt injection vulnerabilities (and coined the term). He recently described the current state of AI security as "horrifying" on his blog, noting that these prompt injection vulnerabilities remain widespread "almost three years after we first started talking about them."

In a prescient warning from September 2022, Willison wrote that "there may be systems that should not be built at all until we have a robust solution." His recent assessment in the present? "It looks like we built them anyway!"

Original Submission

wirelessduck writes:

https://joel.drapper.me/p/rubygems-takeover/

Ruby Central recently took over a collection of open source projects from their maintainers without their consent. News of the takeover was first broken by Ellen on 19 September.

I have spoken to about a dozen people directly involved in the events, and seen a recording of a key meeting between Ruby Gems maintainers and Ruby Central, to uncover what went on.

https://narrativ.es/@janl/115258495596221725

Okay so this was a hostile takeover. The Ruby community needs to get their house in order.

And one more note, I assume it is implied in the write up, but you might now know: DHH is on the board of directors of Shopify. He exerts tremendous organisational and financial power.

It's hilarious he's threatened by three devs with a hobby project and is willing to burn his community's reputation over it.

Original Submission

looorg writes:

They finally came up with a word for it. "Workslop". To much AI usage among (co-)workers is leading to "workslop". Where there is to much AI production that doesn't turn out to be very valuable or productive. It looks fine at a first glance but has produced nothing of value or solved any problems. It just looks fine. All shiny surface, but nothing actual.

workslop is "AI generated work content that masquerades as good work, but lacks the substance to meaningfully advance a given task."

AI promised to revolutionize productivity. Instead, 'workslop' is a giant time suck and the scourge of the 21st century office, Stanford warns

A benefits manager said of one AI-sourced document a colleague sent her, "It was annoying and frustrating to waste time trying to sort out something that should have been very straightforward."

So while companies may be spending hundreds of millions on AI software to create efficiencies and boost productivity, and encouraging employees to use it liberally, they may also be injecting friction into their operations.

The researchers say that "lazy" AI-generated work is not only slowing people down, it's also leading to employees losing respect for each other. After receiving workslop, staffers said they saw the peers behind it as less creative and less trustworthy.

"The insidious effect of workslop is that it shifts the burden of the work downstream, requiring the receiver to interpret, correct, or redo the work," they write.

So shit literally flows downwards then?

https://hbr.org/2025/09/ai-generated-workslop-is-destroying-productivity
https://techcrunch.com/2025/09/27/beware-coworkers-who-produce-ai-generated-workslop/
https://fortune.com/2025/09/23/ai-workslop-workshop-workplace-communication/
https://edition.cnn.com/2025/09/26/business/ai-workslop-nightcap

Original Submission