Slashdot

"About 59% of TikTok videos served to a new account's For You feed are AI slop," writes Search Engine Journal, "according to a report from Kapwing, the video creation tool company. That's roughly three times the rate Kapwing found on YouTube." The company manually reviewed over 10,000 TikTok videos across 20 categories and ran a separate fresh-account test, counting AI-generated content in the first 500 For You videos. Kapwing ran the same fresh-account test on YouTube and found that 104 of the first 500 Shorts, or 21%, were AI slop. On TikTok, 294 of 500 For You videos hit that threshold... Of the 2,000 videos Kapwing reviewed in TikTok's Kids category, 57% were AI slop. That was the highest rate of any category in the analysis. The highest-rate tag was #cartoonkids, where 97 of 100 featured videos were AI-generated. Tags like #cartoons and #babysong both reached 83%, and #forkids came in at 79%. After Kids, the next highest AI slop rates were in Science and Education (35%), Health (33%), and History (33%). All three are categories where visual illustration and voiceover narration make up much of the content. On the other end, categories where on-camera presence or physical demonstration are central had the lowest rates. Fashion came in at 1.3%, Music at 1.5%, and Fitness at 1.6%. The article notes that by last November, TikTok "had already labeled 1.3 billion videos as AI-generated, according to the report."

Read more of this story at Slashdot.

The blog Linuxiac reports: A new systemd fork has appeared with a specific purpose: removing systemd's recently added support for storing a user's birth date in JSON user records. The fork, called Liberated systemd, published its first tagged release as v261 shortly after the official systemd 261 release. In other words, the fork follows upstream systemd while reverting the change that added the new optional birthDate field. Importantly, this is not a new init system, a wider redesign of systemd, or a general-purpose alternative to the upstream project. Its stated purpose is to remain close to upstream systemd while removing what the author describes as "surveillance enablement"... The author recommends testing the fork in a virtual machine before using it on real hardware and warns nightly builds are more likely to be unstable than named releases.

Read more of this story at Slashdot.

"There's a revolution in battery technology hiding in plain sight," reports The Wall Street Journal. "The 3-D printing of batteries has the potential to put energy storage inside any device. "This will enable lightweight and long-lasting consumer gadgets, long-range military drones and even nanoscale robots." Almost all the innovations we regularly hear about — from cheaper, tougher electric-vehicle batteries to "Holy Grail" solid-state batteries — are about changing the chemistry of batteries. The promise of battery-tech 3-D printing (aka additive manufacturing) is simple: What if batteries could fill any available space, even structural elements of our gadgets, rather than always taking a rigid shape like a pouch or cylinder? The new approach has obvious appeal. The entire airframe of a drone could be filled with energy storage for increased range. Smartglasses could have sleek battery-packed frames, so they look like everyday eyewear rather than "Revenge of the Nerds" props. One of the biggest advantages of 3-D printing is that it works with any battery, regardless of its cell chemistry. It could advance today's lithium-ion as well as emerging sodium-ion and solid-state tech... Some [startups] are trying to use 3-D printing to create efficiencies in existing battery manufacturing systems. A brave handful of startups are pursuing radical new designs and approaches. They're starting with defense applications, where cost and scale are less of an issue... At Silicon Valley-based Sakuu... [r]ather than trying to 3-D-print whole batteries, the company is working on replacing one of battery manufacturing's biggest pain points, says Arwed Niestroj, Sakuu's chief operating officer, who is also a nuclear physicist and former head of Mercedes-Benz Research & Development North America. Existing battery assembly lines include football-field-long ovens for drying layers of material that have been dissolved in solvents. This requires a huge amount of energy and is a significant contributor to manufacturing costs, a big reason EV batteries aren't cheaper. Sakuu's process, under development for years, uses additive manufacturing to lay down key battery components without solvents, eliminating the need for ovens, says Niestroj. Sakuu is currently working to commercialize this tech with a major battery manufacturer...

Read more of this story at Slashdot.

Electrek reports: Tesla wants to sell modular AI data center hardware, according to a new trademark application for a product called "Megapod." The filing describes a complete, self-contained computing system for AI workloads... Tesla filed the "Megapod" trademark (serial number 99893717) with the U.S. Patent and Trademark Office this month, through its longtime IP counsel. It's an intent-to-use application, meaning Tesla is claiming the name for a product it hasn't launched yet. The goods-and-services description is unusually specific for a trademark. Megapod covers "modular data center hardware systems for artificial intelligence computing, comprised of computer servers, computer hardware for artificial intelligence data processing, networking equipment, power distribution units, and cooling systems." It also covers "self-contained modular computing hardware systems for artificial intelligence workloads," integrated platforms sold as a single unit — an enclosure bundling compute, power distribution, and cooling — and downloadable software to monitor, manage, and optimize those systems. In plain terms: Tesla wants to sell a turnkey AI data center building block. Not a battery, not a chip on its own, but the full rack-and-room of servers, networking, power, and cooling that AI training and inference run on. Tesla's offering would have to compete with Nvidia's liquid-cooled, rack-scale systems that simulates a giant GPU, the article points out. But "The bigger issue is that Tesla has no merchant compute-hardware business to build on." Tesla's own AI training cluster, Cortex at Gigafactory Texas, runs on roughly 67,000 Nvidia H100-equivalent GPUs. In other words, Tesla is one of Nvidia's customers, not a competitor selling alternative hardware... Where Tesla does have a real AI-data-center business is power, not compute. Its Megapack and new Megablock energy storage products are selling into AI data centers as grid buffers — Musk's own xAI has bought roughly $1 billion of Megapacks to keep its training runs powered. That energy-storage strength is the one credible thread here. A Megapod that bundles Tesla's power electronics, thermal management, and the enclosure — the "shell" around the chips rather than the chips themselves — would at least sit adjacent to a business Tesla actually runs.

Read more of this story at Slashdot.

PC Gamer reports: The UK government is considering an Australia-style ban on social media for under-16s, with Prime Minister Keir Starmer saying that the ban could take effect as soon as spring next year. As for the much nearer future, Science and Technology Secretary Liz Kendall told BBC Breakfast earlier this week, "We will make further statements in July about VPNs and further restrictions." To be clear, no specific restrictions have yet been announced and Kendall sounded somewhat cautious about an outright ban during a parliament debate that took place the same day. "I have commissioned further research about their usage. There are really important issues to balance here," she says. "Many people want to use VPNs for privacy — that is important — but we know that some children use them to get around restrictions. I will come back to that in July in our response to the consultation." So, we'll have to wait until next month for anything definite, but it's hard not to feel like a full ban on VPNs is already on the table. If that does come to pass, more than the contents of my Bluesky inbox will be at stake. Utah in the US has already tried to implement a full VPN ban (though this was postponed until September after Aylo, the parent company of Pornhub, challenged the law in court)... [T]he UK could just be the next domino after Utah, potentially setting off a chain reaction that affects users around the world. The article also argues that age checks can also be a privacy nightmare "with the security breach that exposed the personal info of 70,000 Discord users last year being one case in point." Here's the complete statement from UK Technology Secretary Kendall. "I'll come back in July with a further statement around VPNs but also additional measures that we want to look at, further restrictions on AI chatbots that parents have found very worrying, more about overnight curfews or breaks in doomscrolling for 16- and 17-year-olds."

Read more of this story at Slashdot.

404 Media remembers how a Florida police office looked up his ex-girlfriend's license plate in the Flock automated license plate reader system at least 69 times in 2024 — even searching for her mom's license plate at least 24 times. The police office was charged with stalking and hacking-related offenses, serving one day in prison with five years of probation — but his case "was not a one-off." [Alternate link via Bruce Schneier] Local news reports from around the country repeatedly detail police abusing the Flock surveillance system in order to stalk their partners or ex-partners. The contours of each story are much the same, with the police officer in question using their access to the system to repeatedly track a specific person over the course of weeks or months. The cases highlight the fact that Flock can be used to track the whereabouts of individual people, that police do not get a warrant in order to use the system, and that, if they have access to the system, they have the technical ability to look up any license plate they want for any reason they want. An April study by the civil rights group Institute for Justice found that at least 18 police officers have been caught around the country using Flock to stalk a romantic interest in the last few years; another database, called the ALPR Abuse Library, has documented 20 specific cases of "stalking/targeting" around the country. The known cases of police stalking are almost certainly a vast underreporting of the overall abuse, because they largely include only cases in which the behavior was so egregious that it led to police officers being fired, arrested, or both. Flock told 404 Media that it is "aware of 15 incidents of abuse, each surfaced because of the transparency and accountability features deliberately built into our platform.... There are also 140,000 monthly active users of Flock, so the relatively rare instances of abuse, while obviously wrong and awful, are exactly that — rare," a Flock spokesperson told 404 Media. [One in 10,000.] "Humans are fallible; unlike most tools society provide law enforcement, Flock ensures that in the instances when our technology is misused, the evidence used to hold responsible parties accountable, is right there in our system. We also encourage all our customers to have a usage policy, regular training, and to implement our Audit Assistance tool, which proactively flags unintended use...." But it is also the case that Flock has strenuously fought against lawsuits and potential regulations that are seeking to require police to get a warrant to use the system. And many cases of abuse have not been detected by police departments themselves but by those private citizens, journalists, and stalking victims who have found patterns of abuse in public records files they have obtained from their local police departments. In most cases of Flock-related stalking reviewed by 404 Media, the abuse occurred over the course of months or years, and the victims were subjected to dozens or hundreds of lookups. Other abuse cases have been discovered using the website HaveIBeenFlocked.com, a website that compiles Flock searches released via public records requests and turns them into a searchable database. Flock has repeatedly tried to get that website taken down, as we have previously reported.

Read more of this story at Slashdot.

Tech Times reports: Linux 7.2's merge window closed out a cleanup campaign on Friday that most kernel developers had stopped expecting to see end: the complete removal of strncpy(), a C string-copy function that the kernel's own documentation labels "actively dangerous," from every subsystem, driver, and architecture-specific file in the kernel source tree. The merge landed June 20, 2026. After around 362 commits spread across six years of incremental work, no call site using the function remained, and the function itself — including the last per-CPU-architecture optimized implementations — was struck from the source. The removal matters beyond housekeeping. strncpy() is a persistent source of a specific class of memory error: kernel buffers that contain sensitive data can leak bytes past an unterminated string boundary, a pattern that enables memory disclosure vulnerabilities. Eliminating the function from the tree removes that entire class from the kernel's attack surface — and, critically, makes strncpy() unavailable to any future contributor, turning a best-practice suggestion into an enforced policy. Phoronix notes it's replaced by five different functions: In place of strncpy, Linux kernel code should use strscpy() for NUL terminated destinations, strscpy_pad() for NUl-terminated destinations with zero-padding, strtomem_pad() for non-NUL-terminated fixed-width fields, memcpy_and_pad() for bounded copies with explicit padding, or memcpy() for known-length memory copies. "The reason five functions were needed," explains Tech Times, "is that different parts of the kernel were using strncpy() for five semantically distinct memory operations — each with a different intent, different termination requirement, and different padding behavior. " The original function obscured all of those differences under a single ambiguous name. The 362-commit campaign to replace it was, in effect, a codebase-wide audit that forced every call site to declare its actual intent in code That is an engineering outcome with lasting value: the kernel's string-handling semantics are now explicit where they were previously implicit, and future maintainers can read a function name and understand what a copy operation actually does.

Read more of this story at Slashdot.

NBC News reports: A group of companies that specialize in tracking international shipments of sensitive technologies is backing a Capitol Hill bill that would require America's most powerful AI chips to incorporate stronger security mechanisms aimed at preventing the chips from reaching China and other adversaries. The letter, signed by six companies, says the Chip Security Act (CSA) would increase American chip companies' competitiveness and close key loopholes in the U.S. export control regime. The move clashes with claims from semiconductor lobbying groups that the requirements would constrain America's booming chip industry. Sent to congressional leadership Thursday morning and seen by NBC News, the dispatch instead argues that more robust security verification would assure chip customers and manufacturers that they are abiding by sensitive restrictions on chip sales. The companies argue that the boosted confidence will "lead to increased sales, faster export approvals, larger transactions, greater access to new markets, and more expansive chip deals." Despite U.S. export control laws banning sales of advanced AI chips to certain countries, including China, loopholes in current requirements have allowed billions of dollars' worth of America's best AI chips to be sold to entities in third-party countries that can then forward them to China. In just one case in March, the Justice Department charged three people with conspiring to forward $2.5 billion of AI chips to China. The CSA aims to address those loopholes, mandating that chip exporters better track where advanced chips are sent, via either bespoke location-verification hardware or software that can run on existing hardware. That, bill proponents claim, would ensure that sensitive chips could be sold to countries like Malaysia or Indonesia without fear of further transfer to China... Experts say that because chips perform the advanced computations required for frontier AI systems, cutting off access to the chips is crucial to prevent geopolitical rivals from using AI systems for military or economic purposes.

Read more of this story at Slashdot.

While the Rust Foundation has a Security Initiative to protect its ecosystem, "the threats have expanded," they announced this week, "and so has the kind of help maintainers need." Much of this comes back to a single shift: Automated tooling (much of it now built on large language models) has gotten good enough to surface real vulnerabilities in open source code quickly and at scale. That is useful, and several large Rust projects have already received and fixed credible issues found this way. The same tooling has also made it trivial to generate vulnerability reports that look plausible and are worthless. Maintainers across the ecosystem are losing real hours sorting these from the reports that matter, and the noise tends to bury the signal. So, with funding from the Alpha-Omega Project, the Rust Foundation is bringing on a full-time AI Security Engineer in Residence dedicated to the Rust ecosystem. This position is being funded with part of the $12.5M in open source security funding that the Linux Foundation announced in March. The role exists to take pressure off maintainers. The person in this position will use a mix of human-led and AI-assisted methods to proactively review Rust itself and the crates the ecosystem leans on most and help us separate real, exploitable issues from false positives and low-signal noise before anything reaches a maintainer... This role will run full-time for six months to start, with room to extend depending on what we learn and the funding available. Methods, playbooks, and prompts will be documented so the work doesn't end with the contract. We are grateful that Rust is not embarking on this work in isolation. Several other ecosystems have received parallel Alpha-Omega grants for the same kind of work (e.g., the PHP Foundation and the Drupal Association) and we plan to share tooling, triage practices, and what we learn rather than duplicating work A statement from Rust's new AI Security Engineer in Residence acknowledges that "One of our next challenges is the wave of bugs discovered by the next generation of AI-powered developer tools."

Read more of this story at Slashdot.

This week the Ubuntu desktop's director of engineering announced they're bringing speech-to-text dictation to Ubuntu Desktop, aiming for an experience "that feels like a natural part of the desktop while respecting user privacy and running entirely on local hardware." "Speech recognition has become a common feature on modern platforms, and we think it should be a first-class experience on Ubuntu Desktop as well." More details from the blog It's FOSS: For Ubuntu 26.10, the initial version of Myna is expected to be a desktop dictation tool built around GNOME on Wayland with a push-to-talk mechanism gatekeeping when your microphone accepts input. Using it means holding a hotkey, speaking, and letting go. A small activity indicator shows while it is listening, and the transcribed text lands wherever the cursor was sitting when dictation started. Recognition itself happens inside a sandboxed component called the Canonical Inference Snap, while a Speech Orchestrator manages the session and an Audio Adapter handles whatever the microphone picks up, denoising and chunking it before it ever reaches the model... Speech recognition will happen locally, and an internet connection is not needed once the appropriate model is installed... The audio data won't be sticking around either, being stored in a small in-memory buffer that gets discarded the moment the session ends. Features like dictation into password fields, wake words, continuous listening, voice assistants, voice commands, translation, speaker identification, and automatic language detection are all off the table... You should also know that Canonical is looking for feedback before the specs for Myna are finalized, especially from people who already rely on dictation or assistive tools on Linux.

Read more of this story at Slashdot.

"A grassroots movement is forming among everyday tech workers who are demanding their companies develop and deploy AI responsibly," reports TechCrunch. Hoping to leverage that discontent is a new super PAC called the Guardrails Alliance. The New York Times reports that it launched Thursday with backers that included tech employees and labor unions: Guardrails positions itself as a populist political movement that runs on small donations from people in the trenches of the AI boom. The PAC has about $5 million at its disposal today and planGuardrails will buy ads to support Alex Bores, a New York congressional candidate who became Leading the Future's first target and is running in the primaries next week. s to raise $15 million this cycle — small potatoes compared to deep-pocketed adversaries like Leading the Future, which has more than $100 million from tech leaders like OpenAI president Greg Brockman... "This is not about matching [Leading the Future] dollar for dollar," [said the super PAC's co-founder, political operative Shaunna Thomas]. "What this vehicle is meant to do is be a political home for people who are concerned about the way the anti-regulation AI tech sector is trying to manipulate elections." Meanwhile a former Netflix and Warner Bros. executive has launched the Alliance for Responsible Innovation in the Arts & Media, reports Variety, calling it an AI-focused content coalition that says it's dedicated to supporting "responsible and sustainable AI innovation and the importance of human creativity." The initial members of the coalition, announced Monday, include Disney, the New York Times, Adobe, Condé Nast, the Financial Times, ITV, Advance, BBC, Cambridge University Press & Assessment, U.K. publisher Reach and Wiley. Many of the coalition's members have either struck deals with AI companies or are developing their own AI tools... The group plans to argue for legal and policy guardrails around AI's usage, with its funding directed towards analyses, tools and services focused on advancing those initiatives... One of the group's launch advisers is Damian Collins, OBE, who previously served as the U.K. Parliamentary Under-Secretary of State in the Department for Science, Innovation and Technology under prime ministers Boris Johnson and Liz Truss. "Using AI to break the law can never be an acceptable excuse," he said in a statement. "Laws around personal safety, intellectual property and financial crime still apply in the age of AI. This is why ARIAM has been created and why I'm proud to working with this necessary initiative."

Read more of this story at Slashdot.

An anonymous reader shared this report from the Associated Press: Officials in Kansas City, Missouri, are preparing to equip cameras on some public buses with facial recognition software capable of identifying passengers who appear on a list of banned riders or missing persons. Supporters and opponents alike view the effort as a major litmus test for tapping the AI-powered software on a U.S. public transportation system, positioning Kansas City as the latest epicenter of a fierce debate over whether the safety benefits of artificial intelligence are worth the privacy costs. "The idea of running face recognition on a camera that is pointed on live spaces in public is a line that until recently has never really been crossed in the last 25 years," said Jay Stanley, senior policy analyst for the Project on Speech, Privacy and Technology at the American Civil Liberties Union. The state of Missouri declined to help fund the project as expected due to concerns with the facial recognition component. Still, the city is pushing ahead with local and federal money, said Tyler Means, chief mobility and strategy officer at the Kansas City Transportation Authority. "Privacy is always a tricky thing," Means said. "We've always had cameras on our buses. It's just new technology. I think in time it'll smooth over and people will realize, 'Well, it didn't really feel any different'...." Images captured by cameras aboard the buses would immediately be checked against any active alerts, generated when a missing person, banned rider or someone on a law enforcement watch list designated by the transportation authority is identified... After the buses return to the depot, the transportation authority would archive the regular video footage on a local server for up to five years. The company partnering with Kansas City to run the cameras "started using live facial recognition years ago to alert nursing homes when residents left the building," according to the article, and then "brought the technology to correctional institutions and schools." But this is its first attempt at bringing its cameras onto public transportation. The article also includes this quote from Will Owen, communications director for the Surveillance Technology Oversight Project. "City residents should not be guinea pigs for transit systems to test Silicon Valley's latest unproven, biased surveillance tech."

Read more of this story at Slashdot.

In January a college student posted a video showing him winning $100,000 on Polymarket — one of 145 that appeared to show bets adding up to almost $410,000, reports the Wall Street Journal. "But none of those bets were real." Instead its creator was "one of dozens of mostly college-age creators Polymarket paid to film themselves making fake trades and sometimes scoring fake wins," the Journal reports, citing interviews with the creators an an analysis of more than 1,100 of their videos: Polymarket built near-perfect copies of its website, then instructed creators to make simulated trades on those dummy sites and hide that they were being paid by Polymarket. To get the videos to go viral, Polymarket has recruited a social-media army to copy and re-post creators' footage. Though the New York-based company has been banned from offering its primary crypto platform in the U.S. since 2022, the social-media creators are paid to specifically target U.S. users, who can still access the site with a virtual private network... Polymarket hired and worked closely with a marketing contractor to promote the site. In a message reviewed by the Journal, that contractor told its social-media army to repost content made by 10 Polymarket creators in particular... These creators didn't initially identify themselves as paid by Polymarket, although one offered a $20 bonus code in his social-media bio... The company instructed creators not to disclose they are paid, according to creators who have worked with the company. They said the pay often added up to $2,000 to $3,000 a month... A handful of videos the Journal reviewed also contained short glimpses of URLs indicating the sites were test environments for Polymarket engineers... Creators said they send the finished videos to Polymarket for review. If a video isn't engaging enough, or if it bears obvious signs of being faked, Polymarket will ask for the videos to be reshot, the creators said... Polymarket sends creators bullet-point guidance on what to say, according to creators who have worked with the company and a recruiting website... Polymarket's viral clipping campaign racked up more than 140 million views on TikTok, YouTube and Instagram, according to the analytics provider Tubular... Internal materials show that Polymarket and Virality promote videos showing how easy it is to conduct insider trades on the platform. Polymarket has paid clippers to promote at least 19 videos discussing opportunities to use inside information or other tactics to manipulate markets. America's advertising laws "require people who are paid to endorse a product to disclose their ties," the article notes, "although there is some gray area about what's permitted." (After the Journal's investigation, the creators started adding "@polymarket partner" to their bios, the article points out._ And when asked for a comment, Polymarket "said it plans to conduct a comprehensive audit of active promotional content."

Read more of this story at Slashdot.

The gaming news site Aftermath reports: Four gamers are suing Sony Interactive Entertainment for allegedly breaking a California law that requires digital storefronts selling games to make it clear people are buying licenses, not actually owning the games. Sony Interactive Entertainment's PlayStation store uses language like "Buy Now" and "Confirm Purchase," lawyers wrote in a complaint filed on Thursday... "In reality, consumers who 'purchase' digital games through PlayStation do not obtain ownership of those products," lawyers wrote. "Instead, PlayStation grants only a limited, revocable license to access the software, subject to multiple restrictions contained in a separate Software Product License Agreement".... [T]he PlayStation store does have a disclosure. Above the "Confirm Purchase" button, there's a note: "By selecting [Confirm Purchase], you agree to complete the purchase in accordance with the PlayStation Terms of Service before using this content. You further acknowledge that your purchase of this digital product amounts to a license subject to the Software Product License Agreement." These four gamers aren't satisfied with that; they said in the complaint that it's too small, and that "a reasonable customer completing a purchase would not necessarily notice this disclosure." "It's a proposed class action complaint, meaning the group of four gamers is asking a judge to grant them class action status."

Read more of this story at Slashdot.

The Wall Street Journal reports on internet-connected devices — and how every year millions of them "can contain a secret digital backdoor that opens up access to your home internet, so that anyone... can surf the web as if they were you." (And this is especially true for "knockoffs that you buy online"...) In a video report this week they tested two digital picture frames from Amazon and three streaming devices from Walmart "because we heard that they often ship with backdoor software used in cyberattacks. Security experts believe manufacturers are being paid to add this malware, but many people also get tricked into downloading the software onto their phones or computers... Within minutes of turning the devices on, there was a surge of internet traffic... Visits to gambling, porn, cryptocurrency and loads of other sketchy web sites started pouring in from users around the world." (And remote visitors also tried to access Outlook and Gmail accounts...) Residential proxy companies even rent out access to "tens of millions of home networks around the world," according to the report. "But the problem is actually worse than that. Hackers figured out a way to seize control of these backdoors, and they started taking over these residential networks. Last month authorities arrested a 23-year-old Ottawa man, saying he'd taken control of more than a million devices to launch some of the largest cyberattacks anyone had ever seen.." After a couple months the Journal's reporter collected logs of all the traffic, and sent it to an investigator at Comcast, who said both were conducting DDoS attacks. But estimate for the number of infected devices are as low as tens of millions or as high 500 million-plus. "We've seen nation state attacks launched through these kind of endpoints, which means your device sitting in your house is part of a nation state attack against another nation state... We've seen ad fraud, we've seen ticket scalping, we've seen financial fraud." But more importantly, "We have seen some of the largest computer attacks — meaning computers attacking other computers at human request — ever recorded in our digital history in the last several months." At cybersecurity conferences, some are warning "there are much larger ones on the horizon if we don't get a hold of this problem." The company making the picture frame "couldn't be reached for comment," while Amazon said it's been out of stock since last year. Both Amazon and Walmart said they take action when they confirm malware on a third-party product.

Read more of this story at Slashdot.

This week OpenAI announced a 750-task test to to measure "whether AI systems can support realistic life science research tasks, not just answer biology questions." But while OpenAI's top-performing GPT-Rosalind model led the rankings, Slashdot reader BrianFagioli notes that "it achieved a pass rate of just 36.1 percent, failing nearly two-thirds of benchmark tasks." Nerds.xyz points out that means "the best-performing model failed nearly two-thirds of the benchmark's tasks." The benchmark also revealed a familiar weakness. AI systems generally perform better when everything is presented as text. Once they are forced to work with supporting documents, figures, or complex datasets, performance drops noticeably. GPT-Rosalind's pass rate fell from 45.1 percent on text-only tasks to 28.1 percent on tasks involving artifacts or URLs. To be fair, the benchmark is not intended to suggest AI is useless in research. Quite the opposite. OpenAI found that models are becoming increasingly capable of scientific communication, evidence synthesis, and translating research findings into practical explanations. Those are valuable skills, particularly for researchers drowning in information. But LifeSciBench serves as a useful reminder that today's AI systems are still far from autonomous scientists. They can help. They can assist. They can sometimes provide surprisingly useful insights. What they cannot reliably do, however, is replace the expertise, judgment, and skepticism that real scientific research requires.

Read more of this story at Slashdot.

Long-time Slashdot reader smooth wombat writes: Alan Turing, one of the more famous people who worked at Bletchley Park to decipher the German Enigma coding machine, was also working on a separate project. His private papers, known as the Bayley papers for his assistant Donald Bayley who held onto the papers until his death in 2020, reveal Turning had produced a working model of a portable voice encryption device. He even demonstrated it by using a Winston Churchill speech recording. "Weighing just 39 kg, including its power pack," Jack Copeland wrote in an article for IEEE Spectrum, "Delilah would be at home in a truck, a trench, or a large backpack." More from Popular Mechanics: Turingâ(TM)s work at Bletchley Park actually informed the Delilah experimentation he was doing at Hanslope Park, and not just because he used Red Forms, the Army-issue sheets Hanslope staffers were meant to use to alert Bletchley staffers to enemy signals, as his personal scrap paper for Delilah experiments. He drew inspiration from one of the German cipher machines they had decoded at Bletchley; not the famed Enigma machine, but rather the SZ42. While the former relied on Morse Code, the latter utilized a 5-bit telegraph code, which Copeland notes âoewas a forerunner of ASCII and Unicode and is still used by some ham radio operators.â The SZ42 produced an obscuring key of telegraph characters, with an identical key produced to both the sender and receiver. If it could be done for text, Turing reasoned it could be done for sound as well... [T]he reason Delilah fell to the wayside of history isnâ(TM)t because it was a failure, but rather because it simply wasnâ(TM)t needed anymore. By the time Turing had built and demonstrated his device, the war was over. What good was a portable voice encryptor if you had no major enemies trying to intercept your calls, the government reasoned. So funding for the project stopped, and Turingâ(TM)s two-year experiment ended with a whimper. Turingâ(TM)s time as an electrical engineer at Hanslope Park became a footnote in his story, if even that.

Read more of this story at Slashdot.

Long-time tech pundit Robert Cringely started his career at the Stanford Artificial Intelligence Lab back in 1978. Last month 73-year-old Cringely explained why his site went on a two-year hiatus — and it's not just because of a heart attack and a stroke last July: Just like everyone else, I've been busy all this time on Artificial Intelligence, founding with two partners a company called 2Brains... The work we were doing together is unfinished, but it's not stopped. The patents are filed, the architecture is documented, and the small team continuing the work includes me. Cringely's first piece made the cast that "the trillion-dollar bet the AI industry is making right now may be wrong, and that there's an architectural alternative we've patented and built." In Machines of Loving Grace, Amodei made the case that scaling compute would eventually solve essentially every hard problem in artificial intelligence. Buried in that optimism — or maybe not buried, maybe right out in the open — was a quiet absolution. Hallucinations, the embarrassing tendency of these systems to state falsehoods with total confidence, would take care of themselves. Make the models big enough, train them long enough, and the problem dissolves. You don't have to solve it. You just have to wait, and spend. And so the entire AI industry breathed a sigh of relief. I have spent forty years watching this industry, and I know a permission slip when I see one. Because that is what the essay became, whatever Amodei intended. It gave every other person writing nine- and ten-figure checks a reason not to worry about the one thing that should worry them most. The hallucination problem is the difference between a clever toy and a system a hospital or a bank or a court can actually rely on. It is the whole ballgame for enterprise AI. And the prevailing wisdom, blessed from the top, is that you needn't address it directly. Scale will provide... A small company I helped start, 2Brains Inc., set out in 2022 to solve hallucinations — before ChatGPT, before the scaling consensus hardened into received truth, back when the polite assumption was that the problem was simply insurmountable. We did not solve it by waiting for bigger models. We solved it architecturally, by separating the part of the system that generates language from the part that retrieves and verifies facts, and reconciling the two before anything reaches the user. It runs on ordinary processors. It is cheap. And on the industry's own benchmark for this kind of faithfulness, it more than doubles the published baseline, with no fabricated facts in the verified case at all. The article asks whether scaling will, at tremendous cost, eventually reduce hallucinations — or even worse, if the largest companies in the world "are spending a fortune chasing a cure that is not coming." And last week Cringely pitched more advantages for their solution, noting that most prompts aren't even chatbot-level creative prompts — but just requests to retrieve simple data: The reason 2Brains doesn't lie and the reason it's cheap are the same reason. It looks the fact up instead of guessing it — so it cannot fabricate, and the lookup runs on a processor that sips power instead of a chip that gulps it. Trust and thrift are not a trade-off you balance against each other. They fall out of a single design decision. You do not pay extra for the honest version. The honest version is the cheap version. That sentence is the whole company.

Read more of this story at Slashdot.

CNBC reports: Waymo is recalling almost 3,900 robotaxis in the U.S. to fix software issues after some cars drove into freeway construction zones, according to notices filed with the National Highway Traffic Safety Administration. The voluntary recall, the Alphabet-owned company's second in just over a month, followed 13 known incidents where Waymo robotaxis drove into construction zones on freeways in Phoenix, or entered freeway lanes with active construction in the San Francisco area, the filings published Thursday said... A letter posted to the regulator's website... noted that, "Driving through a closed construction zone increases the risk of a crash..." [Waymo said in a statement emailed to CNBC] "We voluntarily restricted freeway operations last month while making improvements, proactively notified state and federal regulators, and decided to file a voluntary software recall with NHTSA. We continue to safely serve riders on surface streets in all the cities where we operate...." The company implemented another voluntary recall in May after some of its robotaxis had driven into flooded zones or standing water. The NHTSA Safety Board also initiated a probe of Waymo after a January incident in which a robotaxi illegally passed a stopped school bus.

Read more of this story at Slashdot.

CNN reports: An unauthorized alert bearing a mysterious message that was sent to cell phones in several states across Brazil on Saturday morning is suspected to be the work of hackers, the Brazilian government said. Devices lit up with the word "misantropi4," an alphanumeric spelling of the Portuguese word "misantropia," which in English translates to "misanthropy". The final letter "a" was substituted with a number '4' — a practice often used by hackers and termed "leetspeak.". The alert — categorized as "extreme" — was initially received in the southern state of Paraná, but a second warning was triggered a few minutes later for cell phones in the major cities of São Paulo and Rio de Janeiro. Brazilian authorities said that the National Civil Defense's warning platform was taken offline after being targeted by a likely hacker attack, and the government is working to restore the tool once all security conditions are reestablished.

Read more of this story at Slashdot.