An anonymous reader shares a report: It's too soon to be talking about the Curse of OpenAI, but we're going to anyway. Since September 10, when Oracle announced a $300 billion deal with the chatbot maker, its stock has shed $315 billion in market value.

OK, yes, it's a gross simplification to just look at market cap. But equivalents to Oracle shares are little changed over the same period (Nasdaq Composite, Microsoft, Dow Jones US Software Index), so the $15 billion loss figure [figure updated with stock price] is not entirely wrong. Oracle's "astonishing quarter" really has cost it nearly as much as one General Motors, or two Kraft Heinz.

Microsoft's Copilot AI assistant in Windows 11 fails to replicate the capabilities shown in the company's TV advertisements. The Verge tested Copilot Vision over a week using the same prompts featured in ads airing during NFL games. When asked to identify a HyperX QuadCast 2S microphone visible in a YouTube video -- a task successfully completed in Microsoft's ad -- Copilot gave multiple incorrect answers. The assistant identified the microphone as a first-generation HyperX QuadCast, then as a Shure SM7b on two other occasions. Copilot couldn't identify the Saturn V rocket from a PowerPoint presentation despite the words "Saturn V" appearing on screen. When asked about a cave image from Microsoft's ad, Copilot gave inconsistent responses.

About a third of the time it provided directions to find the photo in File Explorer. On two occasions it explained how to launch Google Chrome. Four times it offered advice about booking flights to Belize. The cave is Rio Secreto in Playa del Carmen, Mexico. Microsoft spokesperson Blake Manfre said "Copilot Actions on Windows, which can take actions on local files, is not yet available." He described it as "an opt-in experimental feature that will be coming soon to Windows Insiders in Copilot Labs, starting with a narrow set of use cases while we optimize model performance and learn." Copilot cannot toggle basic Windows settings like dark mode. When asked to analyze a benchmark table in Google Sheets, it "constantly misread clear-as-day scores both in the spreadsheet and in the on-page review."

An anonymous reader shares a report: The IRS accessed a database of hundreds of millions of travel records, which show when and where a specific person flew and the credit card they used, without obtaining a warrant, according to a letter signed by a bipartisan group of lawmakers and shared with 404 Media. The country's major airlines, including Delta, United Airlines, American Airlines, and Southwest, funnel customer records to a data broker they co-own called the Airlines Reporting Corporation (ARC), which then sells access to peoples' travel data to government agencies.

The IRS case in the letter is the clearest example yet of how agencies are searching the massive trove of travel data without a search warrant, court order, or similar legal mechanism. Instead, because the data is being sold commercially, agencies are able to simply buy access. In the letter addressed to nine major airlines, the lawmakers urge them to shut down the data selling program. Update: after this piece was published, ARC said it already planned to shut down the program.

"Disclosures made by the IRS to Senator Wyden confirm that it did not follow federal law and its own policies in purchasing airline data from ARC," the letter reads. The letter says the IRS "confirmed that it did not conduct a legal review to determine if the purchase of Americans' travel data requires a warrant."

A federal judge ruled Tuesday that Meta did not illegally stifle competition when it acquired Instagram and WhatsApp. The decision marks Big Tech's first major victory against antitrust enforcement that began during President Donald Trump's first term. The U.S. Federal Trade Commission had sought to force Meta to sell or restructure the platforms to restore competition among social media networks. Meta argued it faced competitive pressure from TikTok, YouTube, and Apple's messaging app.

A majority of global fund managers think companies are overinvesting, as market anxiety grows about the sustainability of the AI spending boom. From a report: A net 20 per cent of fund managers surveyed this month by Bank of America said companies were spending too much on their investments -- the first time this has been a majority view in data running back to 2005. "This jump is driven by concerns over the magnitude and financing of the AI capex boom," said BofA analysts.

The surge in investment to develop AI infrastructure has been a dominant theme in the record rally in US tech stocks this year -- with chipmaker Nvidia becoming the world's first $5tn company last month -- but growing concerns about the sustainability of this spending has caused a pullback on Wall Street in recent weeks.

Google released Gemini 3 on Tuesday, launching its latest AI model with a breakthrough score of 1501 Elo on the LMArena Leaderboard alongside state-of-the-art performance across multiple benchmarks including 91.9% on GPQA Diamond for PhD-level reasoning and 37.5% on Humanity's Last Exam without tool usage. The model is available starting today in the Gemini app, AI Mode in Search for Google AI Pro, Google AI Studio, Vertex AI and the newly launched Google Antigravity agentic development platform. Third-party platforms including Cursor, GitHub, JetBrains, Manus, and Replit are also gaining access.

Separately, Google said AI Overviews now have 2 billion users every month. Gemini app has topped 650 million users per month.

Microsoft has rolled out a new preview build for Windows 11 Insiders in the Dev and Beta Channel this week that introduces a new toggle called 'experimental agentic features' that can be enabled or disabled in the Windows Settings app. From a report: According to Microsoft, this new toggle is designed to "allow agents to use new Windows agentic features." The company says the feature will work with AI-powered apps, which "help you automate everyday tasks -- like organizing files, scheduling meetings, or sending emails -- so you can spend less time on busy work and more time on what matters most. One powerful way apps are implementing AI today is by interacting with your apps and your files, using vision and advanced reasoning to click, type and scroll like a human would."

The setting in the Windows Setting says "When this setting is on, agents can use Windows agentic features." Features such as the recently announced Copilot Actions for Windows feature are going to take advantage of this new experimental agentic feature capability.

Microsoft, Nvidia and OpenAI-rival Anthropic announced strategic partnerships today that will scale Claude on Microsoft Azure and bring up to $15 billion in new investment to the AI startup. Anthropic committed to purchase $30 billion of Azure compute capacity and contract additional capacity up to one gigawatt. Nvidia and Microsoft -- the largest investor in OpenAI -- committed to invest up to $10 billion and up to $5 billion respectively in Anthropic.

Researchers at the University of Vienna extracted phone numbers for 3.5 billion WhatsApp users by systematically checking every possible number through the messaging service's contact discovery feature. The technique yielded profile photos for 57% of those accounts and profile text for 29 percent. The researchers checked roughly 100 million numbers per hour using WhatsApp's browser-based app.

The team warned Meta in April and deleted their data. The company implemented stricter rate-limiting by October to prevent such mass enumeration. Meta called the exposed information "basic publicly available information" and said it found no evidence of malicious exploitation. The vulnerability had been identified before. In 2017, Dutch researcher Loran Kloeze published a blog post detailing the same enumeration technique. Meta responded then that WhatsApp's privacy settings were functioning as designed and denied him a bug bounty reward. The researchers collected 137 million U.S. phone numbers. In India, they found nearly 750 million numbers. They also discovered 2.3 million Chinese numbers and 1.6 million Myanmar numbers, despite WhatsApp being banned in both countries. The researchers analyzed the cryptographic keys and found some accounts used duplicate keys. They speculate this resulted from unauthorized WhatsApp clients rather than a platform flaw.

Every company would be affected if the AI bubble were to burst, the head of Google's parent firm Alphabet has told the BBC. From the report: Speaking exclusively to BBC News, Sundar Pichai said while the growth of artificial intelligence investment had been an "extraordinary moment", there was some "irrationality" in the current AI boom. It comes amid fears in Silicon Valley and beyond of a bubble as the value of AI tech companies has soared in recent months and companies spend big on the burgeoning industry.

Asked whether Google would be immune to the impact of the AI bubble bursting, Mr Pichai said the tech giant could weather that potential storm, but also issued a warning. "I think no company is going to be immune, including us," he said. In a wide-ranging exclusive interview at Google's California headquarters, he also addressed energy needs, slowing down climate targets, UK investment, the accuracy of his AI models, and the effect of the AI revolution on jobs.

An anonymous reader quotes a report from Wired: Startup Valar Atomics said on Monday that it achieved criticality -- an essential nuclear milestone -- with the help of one of the country's top nuclear laboratories. The El Segundo, California-based startup, which last week announced it had secured a $130 million funding round with backing from Palmer Luckey and Palantir CTO Shyam Sankar, claims that it is the first nuclear startup to create a critical fission reaction. It's also, more specifically, the first company in a special Department of Energy pilot program aiming to get at least three startups to criticality by July 4 of next year to announce it had achieved this reaction. The pilot program, which was formed following an executive order President Donald Trump signed in May, has upended US regulation of nuclear startups, allowing companies to reach new milestones like criticality at a rapid pace.

There's a difference between the type of criticality Valar reached this week -- what's known as cold criticality or zero-power criticality -- and what's needed to actually create nuclear power. Nuclear reactors use heat to create power, but in cold criticality, which is used to test a reactor's design and physics, the reaction isn't strong enough to create enough heat to make power. The reactor that reached criticality this week is not actually Valar's own model, but rather a blend of the startup's fuel and technology with key structural components provided by the Los Alamos National Laboratory, one of the DOE's research and development laboratories. The combination reactor builds off a separate fuel test performed last year at the laboratory, using fuel similar to what Valar's reactor will use. "Zero power criticality is a reactor's first heartbeat, proof the physics holds," Valar founder Isaiah Taylor said in a statement. "This moment marks the dawn of a new era in American nuclear engineering, one defined by speed, scale, and private-sector execution with closer federal partnership."

An outage at Cloudflare that began moments ago has knocked many popular websites, including ChatGPT and X, according to user reports. Cloudflare says on its website: "Cloudflare is aware of, and investigating an issue which potentially impacts multiple customers. Further detail will be provided as more information becomes available."

Update: In a statement after the outage was resolved, Cloudflare CTO said: Earlier today we failed our customers and the broader Internet when a problem in Cloudflare network impacted large amounts of traffic that rely on us. The sites, businesses, and organizations that rely on Cloudflare depend on us being available and I apologize for the impact that we caused.

Transparency about what happened matters, and we plan to share a breakdown with more details in a few hours. In short, a latent bug in a service underpinning our bot mitigation capability started to crash after a routine configuration change we made. That cascaded into a broad degradation to our network and other services. This was not an attack.

That issue, impact it caused, and time to resolution is unacceptable. Work is already underway to make sure it does not happen again, but I know it caused real pain today. The trust our customers place in us is what we value the most and we are going to do what it takes to earn that back.

schwit1 shares a report from the New York Times: Pete Kayll, a musclebound veteran of Britain's Royal Marines, had an unusual instruction for the Bitcoin investors gathered in Switzerland in late October. "Just bite your way out," he told them. It was the final day of a weekend-long cryptocurrency convention on the shore of Lake Lugano, near the Italian border. A small group of investors had lined up in a conference room to have their hands bound with plastic zipties. Now they were learning how to get them off. "Your teeth will get through anything," Mr. Kayll advised. "But it will bloody well hurt."

Most people don't go to an international crypto conference expecting to learn how to gnaw through plastic. But after hours of panels devoted to topics like Bitcoin-collateralized loans, these investors were looking for something more practical. They wanted to know what to do if they were grabbed on the street and thrown into the back of a van. Already paranoid about scams, hacks and market turmoil, wealthy crypto investors have lately become terrified about a much graver threat: torture and kidnapping. These threats are known as "wrench attacks," which is a reference to a popular XKCD cartoon where a thief skips the hacking and just uses a wrench to force out the password.

According to the NYT, the best way to stay protected is staying low-profile, minimizing visible signs of wealth, using basic physical security tools, and preparing for self-defense. The report specifically recommends avoiding flashy displays of wealth like luxury watches and cars, watching for honey-traps, using hotel door stoppers, practicing escape techniques such as breaking zip-ties, hiring discreet bodyguards, and relying on panic-button apps like Glok to summon help quickly.

UC Berkeley researchers working with Project CETI discovered that sperm whales produce vowel-like sounds embedded in their click codas, suggesting a far more complex communication system than previously understood. "It was striking just how structured the system was. I've never seen anything like that before with other animals," Begus, a UC Berkeley linguistics professor and the linguistics lead at Project CETI, told SFGATE. "We're showing the world that there's more than meets the eye in sperm whales and that, if one cares to look closely, they're not as alien. We're much more similar to each other than we used to think." SFGATE reports: With the help of a machine-learning model to identify patterns, Begus and his team combed through recordings collected from social units of sperm whales off the coast of the island of Dominica between 2005 and 2018. When they sped up the audio, removing the silences between clicks, they heard new patterns. They found acoustic properties that share similarities with two vowels -- a and i -- and several vowel combinations.

"Before, people were looking just at the timing and the number of clicks exchanged between sperm whales, but now we have to look at the frequencies, too. A whole new set of patterns have appeared," Begus said. "Now, it's one of the most complex non-human communication systems we have observed." [...] Begus said the research only shows how much more we have to learn about whales' style of communicating. He is particularly interested in exploring how the system may differ for whales between regions and how whale babies learn to communicate in this way. Most importantly, he wants to understand the meaning behind the sounds, as a "window into whale thoughts and lives." The research was published in the journal Open Mind.

An anonymous reader quotes a report from the New York Times: For the first time, scientists are tracking the migration of monarch butterflies across much of North America, actively monitoring individual insects on journeys from as far away as Ontario all the way to their overwintering colonies in central Mexico. This long-sought achievement could provide crucial insights into the poorly understood life cycles of hundreds of species of butterflies, bees and other flying insects at a time when many are in steep decline.

The breakthrough is the result of a tiny solar-powered radio tag that weighs just 60 milligrams and sells for $200. Researchers have tagged more than 400 monarchs this year and are now following their journeys on a cellphone app created by the New Jersey-based company that makes the tags, Cellular Tracking Technologies. Most monarchs weigh 500 to 600 milligrams, so each tag-bearing migrator making the transcontinental journey is, by weight, equivalent to a half-raisin carrying three uncooked grains of rice.

Researchers are tracking more than 400 tagged monarch butterflies as they fly toward winter colonies in central Mexico. The maps [in the article] follow six butterflies. [...] Tracking the world's most famous insect migration may also have a big social impact, with monarch lovers able to follow the progress of individual butterflies on the free app, called Project Monarch Science. Many of the butterflies are flying over cities and suburbs where pollinator gardens are increasingly popular. Some tracks could even lead to the discovery of new winter hideaways. "There's nothing that's not amazing about this," said Cheryl Schultz, a butterfly scientist at Washington State University and the senior author of a recent study documenting a 22 percent drop in butterfly abundance in North America over a recent 20-year period. "Now we will have answers that could help us turn the tide for these bugs."