news aggregator

After a five-year hiatus, Cornelis' interconnect returns at 400Gbps, with Ethernet support next

Five years after Intel spun off its Omni-Path interconnect tech into Cornelis Networks, its 400Gbps CN5000 line of switches and NICs is finally ready to do battle with its long-time rival, Nvidia's InfiniBand.…

Services locked down in America's second-largest city

Video Five Waymo robotaxis were torched on Sunday during protests against the US Immigration and Customs Enforcement (ICE) agency's efforts to detain and deport suspected undocumented immigrants.…

According to Bloomberg, several major Chinese AI companies, including Alibaba, ByteDance, and Tencent, have temporarily disabled certain chatbot features during the gaokao college entrance exams to prevent cheating. "Popular AI apps, including Alibaba's Qwen and ByteDance's Doubao, have stopped picture recognition features from responding to questions about test papers, while Tencent's Yuanbao, Moonshot's Kimi have suspended photo-recognition services entirely during exam hours," adds The Verge. From the report: The rigorous multi-day "gaokao" exams are sat by more than 13.3 million Chinese students between June 7-10th, each fighting to secure one of the limited spots at universities across the country. Students are already banned from using devices like phones and laptops during the hours-long tests, so the disabling of AI chatbots serves as an additional safety net to prevent cheating during exam season. When asked to explain the suspension, Bloomberg reports the Yuanbao and Kimi chatbots responded that functions had been disabled "to ensure the fairness of the college entrance examinations." Similarly, the DeepSeek AI tool that went viral earlier this year is also blocking its service during specific hours "to ensure fairness in the college entrance examination,"according to The Guardian. The Guardian notes that the news is being driven by students on the Chinese social media platform Weibo. "The gaokao entrance exam incites fierce competition as it's the only means to secure a college placement in China, driving concerns that students may try to improve their chances with AI tools," notes The Verge.

Read more of this story at Slashdot.

Online catalog gives open science access to data from early universe

A NASA-backed project using observations from the James Webb Space Telescope (JWST) has released more than 1.5 TB of data for open science, offering the largest view deep into the universe available to date.…

Apple unveiled iPadOS 26 at its Worldwide Developer Conference on Monday, introducing what appears to be the most significant productivity overhaul in the tablet operating system's history. The update brings dynamically resizable windows that users can drag by their corners, a menu bar accessible through swipe gestures or cursor movement, and Expose for viewing all open windows in a tiled array. The new windowing system allows users to seamlessly close, minimize, resize, and tile app windows while maintaining the iPad's touch-first interface. When users reopen apps, windows return to their last position and size. The menu bar, a longtime Mac staple, provides access to familiar commands like File, Edit, and View through either touch or trackpad controls. Apple is also enhancing the Files app with resizable columns and collapsible folders, while bringing the Preview app to iPad for the first time with PDF editing capabilities and Apple Pencil support. The update introduces Background Tasks for computationally intensive processes and new audio features including Voice Isolation and Local capture for video calls.

Read more of this story at Slashdot.

An anonymous reader quotes a report from Ars Technica: YouTube videos may be getting a bit more pernicious soon. Google's dominant video platform has spent years removing discriminatory and conspiracy content from its platform in accordance with its usage guidelines, but the site is now reportedly adopting a lighter-touch approach to moderation. A higher bar for content removal will allow more potentially inflammatory content to remain up in the "public interest." [...] Beginning late last year, YouTube began informing moderators they should err on the side of caution when removing videos that are in the public interest. That includes user uploads that discuss issues like elections, race, gender, sexuality, abortion, immigration, and censorship. Previously, YouTube's policy told moderators to remove videos if one-quarter or more of the content violated policies. Now, the exception cutoff has been increased to half. In addition, staff are now told to bring issues to managers if they are uncertain rather than removing the content themselves. "Recognizing that the definition of 'public interest' is always evolving, we update our guidance for these exceptions to reflect the new types of discussion we see on the platform today," YouTube's Nicole Bell told the New York Times. "Our goal remains the same: to protect free expression on YouTube while mitigating egregious harm." Most of the videos hosted on YouTube won't be affected by this change, the company says. "These exceptions apply to a small fraction of the videos on YouTube, but are vital for ensuring important content remains available," a YouTube spokesperson tells Ars. "This practice allows us to prevent, for example, an hours-long news podcast from being removed for showing one short clip of violence."

Read more of this story at Slashdot.

An anonymous reader shares a report: Seagate has announced a new 4TB version of its storage expansion card for the Xbox Series X and S consoles. It's the first time the company has introduced a new capacity since launching 2TB and 512GB versions of the expansion card in late 2021. The 4TB card is available starting today through Seagate's online store and Best Buy for $499.99, but is discounted to $429.99 as part of a limited-time launch promotion. For comparison, the Xbox Series S starts at $379.99, while the Xbox Series X starts at $599.99.

Read more of this story at Slashdot.

AI 'thinking' may just be an illusion

If you are betting on AGI – artificial general intelligence, the point at which AI models rival human cognition – showing up next year, you may want to adjust your timeline.…

Apple today introduced Liquid Glass, a new design language that brings transparency and glass shine effects across macOS, iPadOS, iOS, and its other software platforms. Alan Dye, Apple's VP of human interface, described the update as the company's "broadest design update, ever" and "the first time we're introducing a universal design across our platforms." The design overhaul adds glass-like elements throughout iOS 26, including glass edges that appear when users swipe up on the lock screen and similar transparent effects across system interfaces. The changes represent Apple's most significant departure from the iOS 7 design philosophy that has shaped the mobile operating system for over a decade since 2013, when Apple moved away from skeuomorphism. App developers will need to adjust their applications to accommodate the new visual language.

Read more of this story at Slashdot.

United Natural Foods shut down some of its systems on June 5 after spotting network intruders

North American grocery wholesaler United Natural Foods told regulators that a cyber incident temporarily disrupted operations, including its ability to fulfill customer orders.…

United Natural Foods (UNFI), a major distributor of groceries to Whole Foods and other retailers, said on Monday that it was hit by a cyberattack, warning of disruptions to its ability to fulfill and distribute customer orders. From a report: UNFI said in a Monday filing with the U.S. Securities and Exchange Commission that it became aware of unauthorized access to its IT systems last Thursday, and began shutting down portions of its network. The filing added that the company has "implemented workarounds for certain operations in order to continue servicing its customers where possible," but noted that the intrusion has caused ongoing disruptions to its business operations. The Providence, Rhode Island-based company is one of the largest grocery distributors in North America, selling fresh produce, goods, and food products to more than 30,000 stores and supermarket locations across the U.S. and Canada. UNFI also serves as the "primary distributor" to Whole Foods, the Amazon-owned grocery chain. Last year, the two companies extended their long-running contract until May 2032.

Read more of this story at Slashdot.

Economic uncertainty and the race to AI are pillaging the IT job market

The IT job market in the US is being hit from two sides at once: Companies are grappling with fears of a recession stemming from the Trump administration's erratic tariff policy, while AI is increasingly mopping up entry-level jobs.…

The world's oceans are in worse health than realised, scientists have said today, as they warn that a key measurement shows we are "running out of time" to protect marine ecosystems. From a report: Ocean acidification, often called the "evil twin" of the climate crisis, is caused when carbon dioxide is rapidly absorbed by the ocean, where it reacts with water molecules leading to a fall in the pH level of the seawater. It damages coral reefs and other ocean habitats and, in extreme cases, can dissolve the shells of marine creatures. Until now, ocean acidification had not been deemed to have crossed its "planetary boundary." The planetary boundaries are the natural limits of key global systems -- such as climate, water and wildlife diversity -- beyond which their ability to maintain a healthy planet is in danger of failing. Six of the nine had been crossed already, scientists said last year. However, a new study by the UK's Plymouth Marine Laboratory (PML), the Washington-based National Oceanic and Atmospheric Administration and Oregon State University's Co-operative Institute for Marine Resources Studies found that ocean acidification's "boundary" was also reached about five years ago.

Read more of this story at Slashdot.

Brace yourselves Britain, PM Keir Starmer's challenged his teams: 'show me how they can use AI'

Britain's beefiest supercomputer, Isambard-AI, is set to become fully operational this summer, as the government steps up its strategy to push AI everywhere as a driver for economic recovery.…

Big expensive Moon rockets = good. Science = yeah, whatever

While US President Donald Trump and his former best pal, Elon Musk, were having a very public spat, the US Senate fired back with its response to NASA's proposed budget cuts. Big rockets = good. Science = still bad.…

A cybersecurity researcher was able to figure out the phone number linked to any Google account, information that is usually not public and is often sensitive, according to the researcher, Google, and 404 Media's own tests. From a report: The issue has since been fixed but at the time presented a privacy issue in which even hackers with relatively few resources could have brute forced their way to peoples' personal information. "I think this exploit is pretty bad since it's basically a gold mine for SIM swappers," the independent security researcher who found the issue, who goes by the handle brutecat, wrote in an email. [...] In mid-April, we provided brutecat with one of our personal Gmail addresses in order to test the vulnerability. About six hours later, brutecat replied with the correct and full phone number linked to that account. "Essentially, it's bruting the number," brutecat said of their process. Brute forcing is when a hacker rapidly tries different combinations of digits or characters until finding the ones they're after. Typically that's in the context of finding someone's password, but here brutecat is doing something similar to determine a Google user's phone number. Brutecat said in an email the brute forcing takes around one hour for a U.S. number, or 8 minutes for a UK one. For other countries, it can take less than a minute, they said. In an accompanying video demonstrating the exploit, brutecat explains an attacker needs the target's Google display name. They find this by first transferring ownership of a document from Google's Looker Studio product to the target, the video says. They say they modified the document's name to be millions of characters, which ends up with the target not being notified of the ownership switch. Using some custom code, which they detailed in their write up, brutecat then barrages Google with guesses of the phone number until getting a hit.

Read more of this story at Slashdot.

An anonymous reader shares a report: Meta is in talks to make a multibillion-dollar investment into AI startup Scale AI, according to people familiar with the matter. The financing could exceed $10 billion in value, some of the people said, making it one of the largest private company funding events of all time. [...] Scale AI, whose customers include Microsoft and OpenAI, provides data labeling services to help companies train machine-learning models and has become a key beneficiary of the generative AI boom. The startup was last valued at about $14 billion in 2024, in a funding round that included backing from Meta and Microsoft.

Read more of this story at Slashdot.

Christian Klein sees little benefit from trying to compete with the dominant hyperscalers

The leader of Europe's most valuable company says there is no point in the continent building datacenters to try to compete with US cloud hyperscalers which have already invested in the region.…

Apple researchers have found that state-of-the-art "reasoning" AI models like OpenAI's o3-mini, Gemini (with thinking mode-enabled), Claude 3.7, DeepSeek-R1 face complete performance collapse [PDF] beyond certain complexity thresholds when tested on controllable puzzle environments. The finding raises questions about the true reasoning capabilities of large language models. The study, which examined models using Tower of Hanoi, checker jumping, river crossing, and blocks world puzzles rather than standard mathematical benchmarks, found three distinct performance regimes that contradict conventional assumptions about AI reasoning progress. At low complexity levels, standard language models surprisingly outperformed their reasoning-enhanced counterparts while using fewer computational resources. At medium complexity, reasoning models demonstrated advantages, but both model types experienced complete accuracy collapse at high complexity levels. Most striking was the counterintuitive finding that reasoning models actually reduced their computational effort as problems became more difficult, despite operating well below their token generation limits. Even when researchers provided explicit solution algorithms, requiring only step-by-step execution rather than creative problem-solving, the models' performance failed to improve significantly. The researchers noted fundamental inconsistencies in how models applied learned strategies across different problem scales, with some models successfully handling 100-move sequences in one puzzle type while failing after just five moves in simpler scenarios.

Read more of this story at Slashdot.

If gamers can have a slimline version of the OS, why not IT admins?

Microsoft just demonstrated it can put Windows 11 on a diet if it really wants to, with the announcement of PC gaming handhelds running a slimmed-down version of the operating system under the hood.…