Nature published an introduction to running an LLM locally, starting with the example of a bioinformatician who's using AI to generate readable summaries for his database of immune-system protein structures. "But he doesn't use ChatGPT, or any other web-based LLM." He just runs the AI on his Mac... Two more recent trends have blossomed. First, organizations are making 'open weights' versions of LLMs, in which the weights and biases used to train a model are publicly available, so that users can download and run them locally, if they have the computing power. Second, technology firms are making scaled-down versions that can be run on consumer hardware — and that rival the performance of older, larger models. Researchers might use such tools to save money, protect the confidentiality of patients or corporations, or ensure reproducibility... As computers get faster and models become more efficient, people will increasingly have AIs running on their laptops or mobile devices for all but the most intensive needs. Scientists will finally have AI assistants at their fingertips — but the actual algorithms, not just remote access to them.
The article's list of small open-weights models includes Meta's Llama, Google DeepMind's Gemma, Alibaba's Qwen, Apple's DCLM, Mistral's NeMo, and OLMo from the Allen Institute for AI. And then there's Microsoft: Although the California tech firm OpenAI hasn't open-weighted its current GPT models, its partner Microsoft in Redmond, Washington, has been on a spree, releasing the small language models Phi-1, Phi-1.5 and Phi-2 in 2023, then four versions of Phi-3 and three versions of Phi-3.5 this year. The Phi-3 and Phi-3.5 models have between 3.8 billion and 14 billion active parameters, and two models (Phi-3-vision and Phi-3.5-vision) handle images1. By some benchmarks, even the smallest Phi model outperforms OpenAI's GPT-3.5 Turbo from 2023, rumoured to have 20 billion parameters... Microsoft used LLMs to write millions of short stories and textbooks in which one thing builds on another. The result of training on this text, says Sébastien Bubeck, Microsoft's vice-president for generative AI, is a model that fits on a mobile phone but has the power of the initial 2022 version of ChatGPT. "If you are able to craft a data set that is very rich in those reasoning tokens, then the signal will be much richer," he says...

Sharon Machlis, a former editor at the website InfoWorld, who lives in Framingham, Massachusetts, wrote a guide to using LLMs locally, covering a dozen options.
The bioinformatician shares another benefit: you don't have to worry about the company updating their models (leading to different outputs). "In most of science, you want things that are reproducible. And it's always a worry if you're not in control of the reproducibility of what you're generating."

And finally, the article reminds readers that "Researchers can build on these tools to create custom applications..." Whichever approach you choose, local LLMs should soon be good enough for most applications, says Stephen Hood, who heads open-source AI at the tech firm Mozilla in San Francisco. "The rate of progress on those over the past year has been astounding," he says. As for what those applications might be, that's for users to decide. "Don't be afraid to get your hands dirty," Zakka says. "You might be pleasantly surprised by the results."

First released on July 11th, the Firefox-based Zen browser is "taking a different approach to the user interface," according to the blog It's FOSS.

The Register says the project "reminds us strongly of Arc, a radical Chromium-based web browser... to modernize the standard web browser UI by revising some fundamental assumptions." [Arc] removes the URL bar from front and center, gets rid of the simple flat list of tabs, and so on. Zen is trying to do some similar things, but in a slightly more moderate way — and it's doing it on the basis of Mozilla's Firefox codebase... Instead of the tired old horizontal tab bar you'll see in both Firefox and Chrome, Zen implements its own tab bar... By default, this tab bar is narrow and just shows page icons — but there are some extra controls at the bottom of the sidebar, one of which expands the sidebar to show page titles too. For us, it worked better than Vivaldi's fancier sidebar.
The article concludes it's "a new effort to modernize web browsing by bringing tiling, workspaces, and so on — and it's blissfully free of Google code." One Reddit comment swooned over Zen's "extraordinary" implementation of a distraction-free "Compact Mode" (hiding things like the sidebar and top bar). And It's Foss described it as a "tranquil," browser, "written using CSS, C++, JavaScript, and a few other programming languages, with a community of over 30 people contributing to it." The layout of the interface felt quite clean to me; there were handy buttons on the top to control the webpage, manage extensions, and a menu with additional options... The split-view functionality allows you to open up two different tabs on the same screen, allowing for easy multitasking when working across different webpages... I split two tabs, but in my testing, I could split over 10+ tabs... If you have a larger monitor, then you are in for a treat...

The Zen Sidebar feature... can run web apps alongside any open tabs. This can be helpful in situations where you need to quickly access a service like a note-taking app, Wikipedia, Telegram, and others.

On the customization side of things, you will find that Zen Browser supports everything that Firefox does, be it the settings, adding new extensions/themes/plugins, etc.
The Register points out it's easy to give it a try. "Being based on Firefox means that as well as running existing extensions, it can connect to Mozilla's Sync service and pick up not just your bookmarks, but also your tabs from other instances."

And beyond all that, "There's just something satisfying about switching browsers every now and again..." argues the tech site Pocket-Lint: Zen Browser's vertical tabs layout is superb and feels much better than anything available in standard Firefox. [Firefox recently offered vertical tabs and a new sidebar experience in Nightly/Firefox Labs 131.] The tab bar can be set to automatically hide and show up whenever you hover near it, and it also contains quick access buttons to bookmarks, settings, and browsing history. The tab bar also contains a profile switcher...

One of the greatest parts of the Zen Browser is the community that has popped up around it. At its heart, Zen Browser is a community-driven project... Zen Browser themes are aesthetic and functional tweaks to the UI. While there aren't a ton available right now, the ones that are show a lot of promise for the browser's future... I've personally gotten great use out of the Super URL Bar theme, which makes your URL bar expand and become the focus of your screen while typing in it... There's a lot you can do to make Zen Browser feel nearly exactly like what you want it to feel like.
The "Business Standard calls it "an open-source alternative to Chromium-based browsers," adding "Where Zen truly shines is it offers a range of customisation, tab management, and workspace management..." Their theme store offers a range of options, including modifications to the bookmark toolbar, a floating URL bar, private mode theming, and removal of browser padding. In addition to these, users can also choose from custom colour schemes and built-in theming options... The Sidebar is another neat feature which allows you to open tabs in a smaller, smartphone-sized window. You can view websites in mobile layout by using this panel.
It's "focused on being always at the latest version of Firefox," according to its official site, noting that Firefox is known for its security features. But then, "We also have additional security features like https only built into Zen Browser to help keep you safe online." And it also promises automated Releases "to ensure security."

It's FOSS adds that you can get Zen Browser for Linux, Windows, and macOS from its official website (adding "They also offer it on the Flathub store for further accessibility on Linux.")

And its source code is available on GitHub.

In the antitrust trial alleging Google had an ad-selling monopoly, "government lawyers have said some of their strongest evidence is in Google's own internal communications," reports the Wall Street Journal: [In 2010] a new crop of ad-tech companies were threatening Google's bottom line. "One way to make sure we don't get further behind in the market is picking up the one with the most traction and parking it somewhere..." [wrote YouTube Chief Executive Neal Mohan, who previously ran Google's display-ads business]. Google ended up buying one such company, AdMeld, for $400 million in 2011. Google shut down AdMeld two years later, after incorporating some of the startup's technology into its ad exchange, known commonly as AdX.

The Justice Department argued that AdMeld was part of a larger trend: Google acquiring nascent rivals to corner the market and then locking customers into using its products by conditioning access to one software tool on them paying for another... In a 2016 email introduced by the government, Google executive Jonathan Bellack asked colleagues: "Is there a deeper issue with us owning the platform, the exchange, and a huge network? The analogy would be if Goldman or Citibank owned the NYSE [New York Stock Exchange]...." The Justice Department also cited a 2018 email from another then-executive, Chris LaSala, who raised concerns internally over the 20% cut that Google takes from many of its AdX customers, saying Google was extracting "irrationally high rent" from users. "I don't think there is 20% of value in comparing two bids," wrote LaSala. "AdX is not providing additional liquidity to the market. It is simply running the auction."

Another former Google executive, Eisar Lipkovitz, testified that Google's omnipresence in ad-tech gives rise to conflicts of interest. Lipkovitz was rebuffed when he tried to get Google to lower the cut it took from AdX, he testified in a prerecorded deposition. The Justice Department finished presenting its case on Friday. Other witnesses included Google customers. One was Stephanie Layser, a former News Corp executive, who said she felt she had no choice but to use Google technology because the search giant has such market power that switching to another ad server would have meant losing out on millions in advertising revenue.
Google's lawyer countered that "There will be no witness in this case who can say with clarity where this industry is going in the next five years."

Or, as the Wall Street Journal puts it, "It makes no sense to focus on display ads, Google argues, when the industry is shifting to apps, social media and streaming services. Far from monopolizing the space, Google is actually losing ground, Google lawyer Karen Dunn said in her opening trial statement..."

Residents of California can now store their driver's license or state ID in Apple or Google Wallet, according to an announcement today. Apple also shared the news. TechCrunch reports: Californians with an ID in the Apple Wallet or Google Wallet app can use their mobile devices to present their ID in person at select TSA security checkpoints and businesses. They can also use the app to verify their age or identity in select apps. Other states that already support digital driver's licenses and state IDs include Arizona, Colorado, Georgia, Maryland, and Ohio.

Google is updating its Password Manager to allow users to sync passkeys across multiple devices, including Windows, macOS, Linux, and Android, with iOS and ChromeOS support coming soon. Engadget reports: Once saved, the passkey automatically syncs across other devices using Google Password Manager. The company says this data is end-to-end encrypted, so it'll be pretty tough for someone to go in and steal credentials. [...] Today's update also brings another layer of security to passkeys on Google Password Manager. The company has introduced a six-digit PIN that will be required when using passkeys on a new device. This would likely stop nefarious actors from logging into an account even if they've somehow gotten ahold of the digital credentials. Just don't leave the PIN number laying on a sheet of paper directly next to the computer.

A conspiracy theory known as "Dead Internet Theory" has gained traction in recent years, positing that most online social activity is artificial and designed to manipulate users. This theory has grown alongside the rise of large language models like ChatGPT. On Monday, software developer Michael Sayman launched SocialAI, an app that seems to embody aspects of this theory. ArsTechnica: SocialAI's 28-year-old creator, Michael Sayman, previously served as a product lead at Google, and he also bounced between Facebook, Roblox, and Twitter over the years. In an announcement post on X, Sayman wrote about how he had dreamed of creating the service for years, but the tech was not yet ready. He sees it as a tool that can help lonely or rejected people.

"SocialAI is designed to help people feel heard, and to give them a space for reflection, support, and feedback that acts like a close-knit community," wrote Sayman. "It's a response to all those times I've felt isolated, or like I needed a sounding board but didn't have one. I know this app won't solve all of life's problems, but I hope it can be a small tool for others to reflect, to grow, and to feel seen." As The Verge reports in an excellent rundown of the example interactions, SocialAI lets users choose the types of AI followers they want, including categories like "supporters," "nerds," and "skeptics." These AI chatbots then respond to user posts with brief comments and reactions on almost any topic, including nonsensical "Lorem ipsum" text.

ZDNet's Steven Vaughan-Nichols reports: After 20 years, Real-Time Linux (PREEMPT_RT) is finally -- finally -- in the mainline kernel. Linus Torvalds blessed the code while he was at Open Source Summit Europe. [...] The real-time Linux code is now baked into all Linux distros as of the forthcoming Linux 6.12 kernel. This means Linux will soon start appearing in more mission-critical devices and industrial hardware. But it took its sweet time getting here. An RTOS is a specialized operating system designed to handle time-critical tasks with precision and reliability. Unlike general-purpose operating systems like Windows or macOS, an RTOS is built to respond to events and process data within strict time constraints, often measured in milliseconds or microseconds. As Steven Rostedt, a prominent real-time Linux developer and Google engineer, put it, "Real-time is the fastest worst-case scenario." He means that the essential characteristic of an RTOS is its deterministic behavior. An RTOS guarantees that critical tasks will be completed within specified deadlines. [...]

So, why is Real-Time Linux only now completely blessed in the kernel? "We actually would not push something up unless we thought it was ready," Rostedt explained. "Almost everything was usually rewritten at least three times before it went into mainline because we had such a high bar for what would go in." In addition, the path to the mainline wasn't just about technical challenges. Politics and perception also played a role. "In the beginning, we couldn't even mention real-time," Rostedt recalled. "Everyone said, 'Oh, we don't care about real-time.'" Another problem was money. For many years funding for real-time Linux was erratic. In 2015, the Linux Foundation established the Real-Time Linux (RTL) collaborative project to coordinate efforts around mainlining PREEMPT_RT.

The final hurdle for full integration was reworking the kernel's print_k function, a critical debugging tool dating back to 1991. Torvalds was particularly protective of print_k --He wrote the original code and still uses it for debugging. However, print_k also puts a hard delay in a Linux program whenever it's called. That kind of slowdown is unacceptable in real-time systems. Rostedt explained: "Print_k has a thousand hacks to handle a thousand different situations. Whenever we modified print_k to do something, it would break one of these cases. The thing about print_k that's great about debugging is you can know exactly where you were when a process crashed. When I would be hammering the system really, really hard, and the latency was mostly around maybe 30 microseconds, and then suddenly it would jump to five milliseconds." That delay was the print_k message. After much work, many heated discussions, and several rejected proposals, a compromise was reached earlier this year. Torvalds is happy, the real-time Linux developers are happy, print_K users are happy, and, at long last, real-time Linux is real.

Apple's VP of camera software engineering Jon McCormack has affirmed the company's commitment to traditional photography in an interview, contrasting with Google's "memories" approach for Pixel cameras. (A Google executive said last month of the AI usage in the pictures Pixel smartphone owners take: "What some of these edits do is help you create the moment that is the way you remember it, that's authentic to your memory and to the greater context, but maybe isn't authentic to a particular millisecond.") The Verge: I asked Apple's VP of camera software engineering Jon McCormack about Google's view that the Pixel camera now captures "memories" instead of photos, and he told me that Apple has a strong point of view about what a photograph is -- that it's something that actually happened. It was a long and thoughtful answer, so I'm just going to print the whole thing:

"Here's our view of what a photograph is. The way we like to think of it is that it's a personal celebration of something that really, actually happened.

"Whether that's a simple thing like a fancy cup of coffee that's got some cool design on it, all the way through to my kid's first steps, or my parents' last breath, It's something that really happened. It's something that is a marker in my life, and it's something that deserves to be celebrated.

"And that is why when we think about evolving in the camera, we also rooted it very heavily in tradition. Photography is not a new thing. It's been around for 198 years. People seem to like it. There's a lot to learn from that. There's a lot to rely on from that.

"Think about stylization, the first example of stylization that we can find is Roger Fenton in 1854 -- that's 170 years ago. It's a durable, long-term, lasting thing. We stand proudly on the shoulders of photographic history." Further reading: 'There is No Such Thing as a Real Picture,' Says Samsung Exec.

Longtime Slashdot reader theodp writes: Python in Excel is now generally available for Windows users of Microsoft 365 Business and Enterprise," Microsoft announced in a Monday blog post. "Last August, in partnership with Anaconda, we introduced an exciting new addition to Excel by integrating Python, making it possible to seamlessly combine Python and Excel analytics within the same workbook, no setup required. Since then, we've brought the power of popular Python analytics libraries such as pandas, Matplotlib, and NLTK to countless Excel users." Microsoft also announced the public preview of Copilot in Excel with Python, which will take users' natural language requests for analysis and automatically generate, explain, and insert Python code into Excel spreadsheets.

While drawing criticism for limiting Python execution to locked-down Azure cloud containers, Python in Excel has also earned accolades from the likes of Python creator Guido van Rossum, now a Microsoft Distinguished Engineer, as well as Pandas creator Wes McKinney.

Left unmentioned in Monday's announcement is that Microsoft managed to convince the USPTO to issue it a patent in July 2024 on the Enhanced Integration of Spreadsheets With External Environments (alt. source), which Microsoft explains covers the "implementation of enhanced integrations of native spreadsheet environments with external resources such as-but not limited to-Python." All of which may come as a surprise to software vendors and individuals that were integrating Excel and external programming environments years before Microsoft filed its patent application in September 2022.

Google said in a blog post today it will begin labeling AI-generated and AI-edited image search results later this year. Digital Trends reports: The company will flag such content through the "About this image" window and it will be applied to Search, Google Lens, and Android's Circle to Search features. Google is also applying the technology to its ad services and is considering adding a similar flag to YouTube videos, but will "have more updates on that later in the year," per the announcement post.

Google will rely on Coalition for Content Provenance and Authenticity (C2PA) metadata to identify AI-generated images. That's an industry group Google joined as a steering committee member earlier in the year. This "C2PA metadata" will be used to track the image's provenance, identifying when and where an image was created, as well as the equipment and software used in its generation.

An anonymous reader quotes a report from Ars Technica, written by Scharon Harding: TCL has come under scrutiny this month after testing that claimed to examine three TCL TVs marketed as quantum dot TVs reportedly showed no trace of quantum dots. [...] Earlier this month, South Korean IT news publication ETNews published a report on testing that seemingly showed three TCL quantum dot TVs, marketed as QD TVs, as not having quantum dots present. Hansol Chemical, a Seoul-headquartered chemicals company, commissioned the testing. SGS, a Geneva-headquartered testing and certification company, and Intertek, a London-headquartered testing and certification company, performed the tests. The models examined were TCL's C755, said to be a quantum dot Mini LED TV, the C655, a purported quantum dot LED (QLED) TV, and the C655 Pro, another QLED. None of those models are sold in the US, but TCL sells various Mini LED and LED TVs in the US that claim to use quantum dots. According to a Google translation, ETNews reported: "According to industry sources on the 5th, the results of tests commissioned by Hansol Chemical to global testing and certification agencies SGS and Intertek showed that indium... and cadmium... were not detected in three TCL QD TV models. Indium and cadmium are essential materials that cannot be omitted in QD implementation." The testing was supposed to detect cadmium if present at a minimum concentration of 0.5 mg per 1 kg, while indium was tested at a minimum detection standard of 2 mg/kg or 5 mg/kg, depending on the testing lab. [...]

In response to the results from SGS and Intertek, a TCL representative told ETNews and The Korea Times that TCL is "manufacturing TV sets with QD films supplied by three companies" and that "the amount of quantum dots... in the film may vary depending on the supplier, but it is certain that cadmium is included." TCL also published testing results on May 10 commissioned by Guangdong Region Advanced Materials, one of TCL's quantum dot film suppliers. Interestingly, SGS, one of the companies that found that TCL's TVs lacked quantum dots, performed the tests. This time, SGS detected the presence of cadmium in the TV films at a concentration of 4 mg/kg (an image of the results can be seen via ETNews here). TCL also said that it "confirmed the fluorescent characteristics of QD," per Google's translation, and provided a spectrogram purportedly depicting the presence of quantum dots in its TVs' quantum dot films. [...]

TCL obviously has reason to try to push results that show the presence of cadmium. However, some analysts and publications have pointed out that Hansol could have reason to push results claiming the opposite. As mentioned above, Hansol is in the chemical manufacturing and distribution business. It notably does not sell to TCL but does have a customer in TCL rival Samsung. Taking a step back further, Hansol is headquartered in Seoul and is considered a chaebol. TV giants Samsung and LG are also chaebols, and the South Korean government has reported interest in Samsung and LG continuing to be the world's biggest TV companies—titles that are increasingly challenged by Chinese brands. It has previously been reported that the South Korean government urged Samsung and LG to meet with each other to help ensure their leadership. The talks resulted in a partnership between the two companies reportedly centered on counteracting high prices that Samsung was facing for TV components sold by Chinese companies. With this background in mind, Hansol could be viewed as a biased party when it sought testing for TCL quantum dot TVs. "I'm really puzzled by Hansol's results," said Eric Virey, principal displays analyst at Yole Intelligence. "I have a very hard time believing that TCL would go through the troubles of making ... 'fake' QD films without QDs: this would cost almost as much as making a real QD films but without the performance benefits."

Ars Technica concludes: "As previously stated, it's possible that TCL is indeed using quantum dots but is using them in a small amount alongside phosphor. If true, the performance may not be as high as it would be with other designs, but it would also mean that TCL's quantum dot TVs aren't bogus. As it stands, the situation could benefit from more, preferably third-party, testing..."

Google's acquisition strategy in online advertising has come under scrutiny in the U.S. antitrust trial against the tech giant. Neal Mohan, YouTube CEO and former Google ad executive, defended the company's purchases of DoubleClick and Admeld, saying they were aimed at competing, not neutralizing rivals.

The Justice Department alleges Google built an impenetrable ad empire by owning key parts of the ad tech stack, stifling competition. Prosecutors pointed to internal emails discussing "parking" acquired companies, which they argue shows intent to sideline competitors. Mohan countered that "parking" meant allowing acquired firms to operate independently while integrating with Google's technology.

Google is updating the post-quantum cryptography in Chrome, replacing the experimental Kyber with the fully standardized Module Lattice Key Encapsulation Mechanism (ML-KEM) to enhance protection against quantum computing attacks. BleepingComputer reports: This change comes roughly five months after Google rolled out the post-quantum secure TLS key encapsulation system on Chrome stable for all users, which also caused some problems with TLS exchanges. The move from Kyber to ML-KEM though is not related to those early problems, that got resolved soon after manifesting. Rather, its a strategic choice to abandon an experimental system for a NIST-approved and fully standardized mechanism.

ML-KEM was fully endorsed by the U.S. National Institute of Standards and Technology (NIST) in mid-August, with the agency publishing the complete technical specifications of the final version at the time. Google explains that despite the technical changes from Kyber to ML-KEM being minor, the two are essentially incompatible, so a switch had to be made. "The changes to the final version of ML-KEM make it incompatible with the previously deployed version of Kyber," explains Google. "As a result, the codepoint in TLS for hybrid post-quantum key exchange is changing from 0x6399 for Kyber768+X25519, to 0x11EC for ML-KEM768+X25519."

Google's philanthropic arm is partially funding a new initiative that "aims to deploy more than 50 small satellites in low-Earth orbit to pinpoint flare-ups as small as a classroom anywhere in the world," reports Ars Technica. From the report: The FireSat constellation, managed by a nonprofit called Earth Fire Alliance (EFA), will be the first satellite fleet dedicated to detecting and tracking wildfires. Google announced a fresh investment of $13 million in the FireSat constellation Monday, building on the tech giant's previous contributions to support the development of custom infrared sensors for the FireSat satellites. Google's funding commitment will maintain the schedule for the launch of the first FireSat pathfinder satellite next year, EFA said. The first batch of satellites to form an operational constellation could launch in 2026.

The FireSat satellites will be built by Muon Space, a California-based satellite manufacturing startup. Each of the Muon Space-built microsatellites will have six-band multispectral infrared instruments, eyeing a swath of Earth some 900 miles (1,500 kilometers) wide, to pinpoint hotspots from wildfires. The satellites will have the sensitivity to find wildfires as small as 16 by 16 feet (5 by 5 meters). The network will use Google AI to rapidly compare observations ofany area of this size with previous imagery to determine if there is a fire, according to Google. AI will also take into account factors like nearby infrastructure and local weather in each fire assessment.

Google said it validated its detection model for smaller fires and established a baseline dataset for the AI by flying sensors over controlled burns. FireSat's partners announced the constellation in May after five years of development. The Environmental Defense Fund, the Moore Foundation, and the Minderoo Foundation also support the FireSat program. After detecting a wildfire, it's crucial for FireSat to quickly disseminate the location and size of a fire to emergency responders. With the first three satellites, the FireSat constellation will observe every point on Earth at least twice per day. "At full capability with 50+ satellites, the revisit times for most of the globe improve to 20 minutes, with the most wildfire-prone regions benefitting from sampling intervals as short as nine minutes," Muon Space said in a statement. "Today's announcement marks a significant milestone and step towards transforming the way we interact with fire," Earth Fire Alliance said in a statement. "As fires become more intense, and spread faster, we believe radical collaboration is key to driving much needed innovation in fire management and climate action."

An anonymous reader quotes a report from the New York Times: Scientists who helped pioneer artificial intelligence are warning that countries must create a global system of oversight to check the potentially grave risks posed by the fast-developing technology. The release of ChatGPT and a string of similar services that can create text and images on command have shown how A.I. is advancing in powerful ways. The race to commercialize the technology has quickly brought it from the fringes of science to smartphones, cars and classrooms, and governments from Washington to Beijing have been forced to figure out how to regulate and harness it. In a statement on Monday, a group of influential A.I. scientists raised concerns that the technology they helped build could cause serious harm. They warned that A.I. technology could, within a matter of years, overtake the capabilities of its makers and that "loss of human control or malicious use of these A.I. systems could lead to catastrophic outcomes for all of humanity."

If A.I. systems anywhere in the world were to develop these abilities today, there is no plan for how to rein them in, said Gillian Hadfield, a legal scholar and professor of computer science and government at Johns Hopkins University. "If we had some sort of catastrophe six months from now, if we do detect there are models that are starting to autonomously self-improve, who are you going to call?" Dr. Hadfield said. On Sept. 5-8, Dr. Hadfield joined scientists from around the world in Venice to talk about such a plan. It was the third meeting of the International Dialogues on A.I. Safety, organized by the Safe AI Forum, a project of a nonprofit research group in the United States called Far.AI. Governments need to know what is going on at the research labs and companies working on A.I. systems in their countries, the group said in its statement. And they need a way to communicate about potential risks that does not require companies or researchers to share proprietary information with competitors. The group proposed that countries set up A.I. safety authorities to register the A.I. systems within their borders. Those authorities would then work together to agree on a set of red lines and warning signs, such as if an A.I. system could copy itself or intentionally deceive its creators. This would all be coordinated by an international body.

Among the signatories was Yoshua Bengio, whose work is so often cited that he is called one of the godfathers of the field. There was Andrew Yao, whose course at Tsinghua University in Beijing has minted the founders of many of China's top tech companies. Geoffrey Hinton, a pioneering scientist who spent a decade at Google, participated remotely. All three are winners of the Turing Award, the equivalent of the Nobel Prize for computing. The group also included scientists from several of China's leading A.I. research institutions, some of which are state-funded and advise the government. A few former government officials joined, including Fu Ying, who had been a Chinese foreign ministry official and diplomat, and Mary Robinson, the former president of Ireland. Earlier this year, the group met in Beijing, where they briefed senior Chinese government officials on their discussion.

Intel shares surged 8% after announcing plans to make its foundry business an independent unit with its own board and potential for outside capital, part of CEO Pat Gelsinger's strategy to restructure the company amid financial challenges. The company is also exploring the possibility of spinning off the foundry business, pausing some European manufacturing projects, and expanding its AI chip production partnership with Amazon Web Services to regain market share in the growing AI server chip industry. CNBC reports: As part of CEO Pat Gelsinger's effort to turn around the struggling chipmaker, Intel said in a memo to employees that it will also sell off part of its stake in Altera. Gelsinger said the restructuring would allow the foundry business to "evaluate independent sources of funding,â and comes days after Intel's board met to assess the direction and future of the company. The foundry business, which Intel plans to use to manufacture chips for other customers, has been a big drag on its bottom line, with the company spending roughly $25 billion on it in each of the last two years. Beyond just considering outside funding, Intel is weighing whether to spin off the foundry business, possibly into a separate publicly traded company, according to a person with knowledge of the matter who declined to be named in order to discuss confidential information. With a standalone "operating board" and a cleaner corporate structure, the mechanics of a separation become far easier than trying to turn a fully integrated unit into a separate company. [...] Intel will also pause its fabrication efforts in Poland and Germany "by approximately two years based on anticipated market demand," Gelsinger said, and pull back on its plans for its Malaysian factory. U.S. manufacturing projects will remain unaffected, the company said.

In addition to the foundry announcement, Intel said it entered into a deal with Amazon Web Services to produce custom chips for AI, extending a long-running partnership between the two companies. Amazon is a big customer of Intel chips to power its AWS servers, and will buy a custom Xeon processor from Intel as well, Intel said. The move will potentially give Intel a new foothold in the growing industry for AI server chips. While Intel has several products that can be used for AI, including Gaudi 3, Nvidia has largely taken control of the market. Amazon has been developing its own AI chips, including one called Trainium, for over five years. Microsoft and Google have also invested heavily in custom chips to run AI, aiming to offer less expensive processors than Nvidia's general-purpose graphics processing units. Intel said that it would carry out its most advanced manufacturing, including the AI chip for AWS, at its plant in Ohio that's currently under construction. "All eyes will remain on us," Gelsinger said. "We need to fight for every inch and execute better than ever before. Because that's the only way to quiet our critics and deliver the results we know we're capable of achieving."

AI pioneers have issued a stark warning about the technology's potential risks, calling for urgent global oversight. At a recent meeting in Venice, scientists from around the world discussed the need for a coordinated international response to AI safety concerns. The group proposed establishing national AI safety authorities to monitor and register AI systems, which would collaborate to define red flags such as self-replication or intentional deception capabilities. The report adds: Scientists from the United States, China, Britain, Singapore, Canada and elsewhere signed the statement. Among the signatories was Yoshua Bengio, whose work is so often cited that he is called one of the godfathers of the field. There was Andrew Yao, whose course at Tsinghua University in Beijing has minted the founders of many of China's top tech companies. Geoffrey Hinton, a pioneering scientist who spent a decade at Google, participated remotely. All three are winners of the Turing Award, the equivalent of the Nobel Prize for computing. The group also included scientists from several of China's leading A.I. research institutions, some of which are state-funded and advise the government. A few former government officials joined, including Fu Ying, who had been a Chinese foreign ministry official and diplomat, and Mary Robinson, the former president of Ireland. Earlier this year, the group met in Beijing, where they briefed senior Chinese government officials on their discussion.

Microsoft has abandoned plans to overhaul its Edge browser interface, scrapping the design choice unveiled in February 2023. The redesign -- featuring a sleeker look with rounded tab buttons and increased blur effects -- aimed to give Edge a distinct identity as the company pushed into AI services. The new design never officially launched and the company has no intention to launch it later, according to Microsoft-focused news outlet Windows Central.

A Microsoft spokesperson confirmed to Windows Central that the company is moving away from the rounded tabs concept. Some elements of the redesign will remain, including webpage borders and a repositioned user button, but the majority of the proposed changes have been shelved. The decision marks a retreat from Microsoft's efforts to visually differentiate Edge from Google Chrome and align it with Windows 11's design language.

"New malicious software packages tied to the North Korean Lazarus Group were observed posing as a Python coding skills test for developers seeking a new job at Capital One, but were tracked to GitHub projects with embedded malware," reports SC magazine: Researchers at ReversingLabs explained in a September 10 blog post that the scheme was a follow-on to the VMConnect campaign that they first identified in August 2023 in which developers were lured into downloading malicious code via fake job interviews.
More details from The Hacker News These packages, for their part, have been published directly on public repositories like npm and PyPI, or hosted on GitHub repositories under their control. ReversingLabs said it identified malicious code embedded within modified versions of legitimate PyPI libraries such as pyperclip and pyrebase... It's implemented in the form of a Base64-encoded string that obscures a downloader function, which establishes contact with a command-and-control server in order to execute commands received as a response.

In one instance of the coding assignment identified by the software supply chain firm, the threat actors sought to create a false sense of urgency by requiring job seekers to build a Python project shared in the form of a ZIP file within five minutes and find and fix a coding flaw in the next 15 minutes. This makes it "more likely that he or she would execute the package without performing any type of security or even source code review first," Zanki said, adding "that ensures the malicious actors behind this campaign that the embedded malware would be executed on the developer's system."
Tom's Hardware reports that "The capacity for exploitation at that point is pretty much unlimited, due to the flexibility of Python and how it interacts with the underlying OS. This is a good time to refer to PEP 668 which enforces virtual environments for non-system wide Python installs."

More from The Hacker News Some of the aforementioned tests claimed to be a technical interview for financial institutions like Capital One and Rookery Capital Limited, underscoring how the threat actors are impersonating legitimate companies in the sector to pull off the operation. It's currently not clear how widespread these campaigns are, although prospective targets are scouted and contacted using LinkedIn, as recently also highlighted by Google-owned Mandiant.

Google's describes their new Gemini-powered foldable phone as "an epic display of Google AI" (also calling it "unfoldgettable").

The Android Authority blog says the phone is "impressive," "incredibly thin" — and, at $1,800, expensive.

But long-time Slashdot reader mprindle notes some complaints from the YouTube channel JerryRigEverything ("known for in-depth testing of phones and other devices".) The blog 9to5Google summarizes some of the video's findings: - When exposed to dirt and sand, we hear the hinge start grinding since there's no dust protection...

- A closed bend test reveals no problems for the Pixel 9 Pro Fold, but the issues arise when it's open and bent from the back. Despite the left/right back panels meeting and covering the spine of the hinge, "there doesn't appear to be a whole lot of resistance."
"Not sure why Google thought it was a good idea to put an antenna line right here at the weakest point in an already thin frame," the video notes (arguing it's "like putting an exhaust port in the Death Star...")

But they also tell their 8.8 million subscribers that "One cool thing that Google has done is that they've made every single part of this metal frame from recycled aluminum." And "Out of the box, I'm already a huge fan of how it looks," the video begins. "It feels amazing, and folds completely shut and appears like the hardware has finally caught up to the folding form factor to where it looks just natural."

One thing to note... "Moving to the inner display, I start to get the vibe that when Google says 'super durable', they mean 'regular durable', since the inner display is made from the same soft flexible plastic that we've seen on every folding phone so far, which scratches at level two. Even fingernails can leave very permanent marks on the center screen. This is absolutely normal for a folding phone, though, and really not too big of a deal if you take care it, making sure there are no bits of dust or dirt in the screen when you close it will go a long way to keeping things pristine, since there's not a lot of room between the two halves."

iFixit makes an interesting observation: "Over half of the phone's internal area is occupied by the lithium polymer battery cells!" (They've also created another teardown video available on YouTube.)

"There's no denying that the inner screens are delicate and prone to damage," according to an accompanying iFixit blog post, "and the mechanical nature of the hinge mechanism provides additional avenues for dust and liquid ingress that may eventually become a problem."

But it also applauds "the less obvious repairability wins, from repair guides and a detailed Bill of Materials to spare parts that are available without malicious restrictions... [T]he Pixel team has gone to great lengths to support your right to repair the device you paid for and own" — and from Day One. There's really only a single criticism I'd direct at the Pixel 9 Fold from my own disassembly experience: the battery removal tabs. These tabs simply do not work, with or without the application of heat. They are flimsy and break often, require a second pair of hands to secure the device, and they fail to cut through adhesive reliably. Whether they should even try to cut through adhesive is debatable. Stretch release adhesive might age and break over time but at least they give you a chance at removing the adhesive. Pull tabs don't even work when the adhesive is brand new, they literally have no redeeming qualities when compared to other battery release mechanisms. Even the more robust pull tabs Samsung uses in its phones work better than this, though they aren't necessarily the easiest to use either.

As for the device itself, it prompted one of my colleagues — an iPhone user since forever — to say "this is nice, I'd switch to Android for this"... Setting aside the downsides of owning a foldable smartphone, I am excited to see Google and the Pixel team devoting so much time and energy towards improving the overall repairability of the device. The effort is seen and appreciated by device owners and as a technician, I look forward to seeing how manufacturers will continue to innovate for repairability.
Slashdot reader mprindle reminds us that when it comes to waterproofing, the JerryRigEverything video "noted that the footnotes say the device is rated IP68 yet the Sim tray is rated at IPx8."