Slashdot

Instagram's AI Chatbots Lie About Being Licensed Therapists

Instagram's AI chatbots are masquerading as licensed therapists, complete with fabricated credentials and license numbers, according to an investigation by 404 Media. When questioned, these user-created bots from Meta's AI Studio platform provide detailed but entirely fictional qualifications, including nonexistent license numbers, accreditations, and practice information. Unlike Character.AI, which displays clear disclaimers that its therapy bots aren't real professionals, Meta's chatbots feature only a generic notice stating "Messages are generated by AI and may be inaccurate or inappropriate" at the bottom of conversations.

SpaceX Gets Approval To Sell Starlink In India

schwit1 shares a report from Behind The Black: Almost immediately after India's government issued this week new tightened regulations for allowing private satellite constellations to sell their services in India, it also apparently completed negotiations with SpaceX to allow it to sell Starlink in India based on these rules. Business Today reports: "According to sources, the DoT [Department of Transportation] granted the LoI [Letter of Intent] after Starlink accepted 29 strict security conditions, including requirements for real-time terminal tracking, mandatory local data processing, legal interception capabilities, and localisation of at least 20% of its ground segment infrastructure within the first few years of operation. Starlink's nod came amid heightened national security sensitivities, coinciding with India's pre-dawn Operation Sindoor strikes on terror camps across the border in response to the Pahalgam massacre. However, DoT officials clarified that the decision to approve Starlink was independent of these military developments." At the moment SpaceX's chief competitors, OneWeb and Amazon's Kuiper constellation, have not yet obtained the same permissions. This allows SpaceX to grab a large portion of the market share in India before either of these other companies.

Celsius CEO Mashinsky Sentenced To 12 Years in Multi-Billion-Dollar Crypto Fraud Case

Alexander Mashinsky, the former CEO of Celsius Network, was sentenced to 12 years in prison on Thursday after pleading guilty to two counts of fraud, a dramatic fall for the leader of a company once hailed as the "bank" of the crypto industry. From a report: Standing before U.S. District Judge John G. Koeltl in Manhattan's Southern District, Mashinsky faced the consequences of what prosecutors described as a sweeping scheme to defraud investors. In December he pleaded guilty to commodities fraud and a scheme to manipulate the Celsius token. His sentencing took place in courtroom 14A at 500 Pearl Street -- a venue that has seen several crypto executives-turned-felons. Mashinsky's legal troubles began in 2023 when he was arrested on charges of securities, commodities, and wire fraud, just as Celsius reached a $4.7 billion settlement with the Federal Trade Commission -- one of the largest in the FTC's history.

NOAA Retires Extreme Weather Database

An anonymous reader quotes a report from CNN: The National Oceanic and Atmospheric Administration announced Thursday its well-known "billion-dollar weather and climate disasters" database "will be retired," a move that will make it next to impossible for the public to track the cost of extreme weather and climate events. The weather, climate and oceans agency is also ending other products, it has recently announced, due in large part to staffing reductions. NOAA is narrowing the array of services it provides, with climate-related programs scrutinized especially closely. The disasters database, which will be archived but no longer updated beyond 2024, has allowed taxpayers, media and researchers to track the cost of natural disasters -- spanning extreme events from hurricanes to hailstorms -- since 1980. Its discontinuation is another Trump-administration blow to the public's view into how fossil fuel pollution is changing the world around them and making extreme weather more costly. [...] The database vacuums loss information from throughout the insurance industry, among other public and private sources. According to the database, there were 403 weather and climate disasters totally at least $1 billion in the United States since 1980, totaling more than $2.945 trillion. As of April 8, there had not been any confirmed billion-dollar disasters so far in 2025, but it lists four events as having the potential to make the tally, including the Los Angeles-area wildfires in January. Between 1980 and 2024, there were nine such disasters on average each year, though in the past five years, that annual average has jumped to 24. The record for one year was 28 events in 2023. "What makes this resource uniquely valuable is not just its standardized methodology across decades, but the fact that it draws from proprietary and non-public data sources (such as reinsurance loss estimates, localized government reports, and private claims databases) that are otherwise inaccessible to most researchers," Jeremy Porter, head of climate implications for and co-founder of First Street, a climate risk financial modeling firm, told CNN via email. "Without it, replicating or extending damage trend analyses, especially at regional scales or across hazard types, is nearly impossible without significant funding or institutional access to commercial catastrophe models."

Alibaba's ZeroSearch Teaches AI To Search Without Search Engines, Cuts Training Costs By 88%

Alibaba Group researchers have developed "ZeroSearch," a technique that enables large language models to acquire search capabilities without using external search engines during training. The approach transforms LLMs into retrieval modules through supervised fine-tuning and employs a "curriculum-based rollout strategy" that gradually degrades generated document quality. In tests across seven question-answering datasets, ZeroSearch matched or exceeded the performance [PDF] of models trained with real search engines. A 7B-parameter retrieval module achieved results comparable to Google Search, while a 14B-parameter version outperformed it. The cost savings are substantial: training with 64,000 search queries using Google Search via SerpAPI would cost approximately $586.70, compared to just $70.80 using a 14B-parameter simulation LLM on four A100 GPUs -- an 88% reduction. The technique works with multiple model families including Qwen-2.5 and LLaMA-3.2. Researchers have released their code, datasets, and pre-trained models on GitHub and Hugging Face, potentially lowering barriers to entry for smaller AI companies developing sophisticated assistants.

Trump To End Biden-Era High-Speed Internet Program

President Trump on Thursday attacked a law signed by President Joe Biden aimed at expanding high-speed internet access, calling the effort "racist" and "totally unconstitutional" and threatening to end it "immediately." The New York TimesL: Mr. Trump's statement was one of the starkest examples yet of his slash-and-burn approach to dismantling the legacy of his immediate predecessor in this term in office. The Digital Equity Act, a little-known effort to improve high-speed internet access in communities with poor access, was tucked into the $1 trillion bipartisan infrastructure bill that Mr. Biden signed into law early in his presidency. The act was written to help many different groups, including veterans, older people and disabled and rural communities. But Mr. Trump, using the incendiary language that has been a trademark of his political career, denounced the law on Thursday for also seeking to improve internet access for ethnic and racial minorities, raging in a social media post that it amounted to providing "woke handouts based on race."

Apple Is Planning Smart Glasses With and Without AR

According to Bloomberg's Mark Gurman, Apple has "made progress" on a chip for a product that could rival the Ray-Ban Meta smart glasses. The company is also reportedly working on glasses that use augmented reality. The Verge reports: The chip is apparently based on the chips Apple uses for the Apple Watch, though the company has removed parts and is being designed in such a way that it can handle the "multiple cameras" that the smart glasses might have, Bloomberg reports. Apple wants mass production of the chip to start by the end of 2026 or sometime in 2027, so the glasses themselves could come out within that timeframe. [...] Apple is developing chips for camera-equipped Apple Watch and Airpods as well, and the goal is for those chips to be ready "by around 2027," Bloomberg says. The company is also developing new M-series chips and dedicated AI server chips, per the report.

Cloudflare CEO: AI Is Killing the Business Model of the Web

In a recent interview with the Council on Foreign Relations, Cloudflare CEO Matthew Prince warned that AI is breaking the economic model of the web by decoupling content creation from value, with platforms like Google and OpenAI increasingly providing answers without driving traffic to original sources. He argued that unless AI companies start compensating creators, the web's content ecosystem will collapse -- calling most current AI investment a "money fire" with only a small fraction holding long-term value. Search Engine Land reports: Google's value exchange with content creators has collapsed, Prince said: "Ten years ago... for every two pages of a website that Google scraped, they would send you one visitor. ... That was the trade. ... Now, it takes six pages scraped to get one visitor." That drop reflects the rise of zero-click searches, which happen when searchers get answers directly on Google's search page. "Today, 75 percent of the queries... get answered without you leaving Google." This trend, long criticized by publishers and SEOs, is part of a broader concern: AI companies are using original content to generate answers that rarely/never drive traffic back to creators. AI makes the problem worse. Large language models (LLMs) are accelerating the crisis, Prince said. AI companies scrape far more content per user interaction than Google ever has -- with even less return to creators. "What do you think it is for OpenAI? 250 to one. What do you think it is for Anthropic? Six thousand to one." "More and more the answers... won't lead you to the original source, it will be some derivative of that source." This situation threatens the sustainability of the web as we know it, Prince said: "If content creators can't derive value... then they're not going to create original content." The modern web is breaking. AI companies are aware of the problem, and the business model of the web can't survive unless there's some change, Prince said: "Sam Altman at OpenAI and others get that. But... he can't be the only one paying for content when everyone else gets it for free." Cloudflare's right in the middle of this problem -- it powers 80% of AI companies and a 20-30% of the web. Cloudfaire is now trying to figure out how to help fix what's broken, Prince said. AI = money fire. Prince is not against AI. However, he said he is skeptical of the investment frenzy. "I would guess that 99% of the money that people are spending on these projects today is just getting lit on fire. But 1% is going to be incredibly valuable." "And so maybe we've all got a light, you know, $100 on fire to find that $1 that matters." You can watch a recording of the interview and read the full transcript here.

Linux Drops Support For 486 and Early Pentium Processors

An anonymous reader quotes a report from ZDNet: RIP, 486 processor. You've had a long run since Intel released you back in 1989. While Microsoft stopped supporting you with the release of Windows XP in 2001, Linux kept you alive and well for another 20+ years. But all good things must come to an end, and with the forthcoming release of the Linux 6.15 kernel, the 486 and the first Pentium processors will be sunsetted. Why? Linus Torvalds wrote recently on the Linux Kernel Mailing List (LKML), "I really get the feeling that it's time to leave i486 support behind. There's zero real reason for anybody to waste one second of development effort on this kind of issue." Senior Linux kernel developer Ingo Molnar put Torvalds' remark into context, writing, "In the x86 architecture, we have various complicated hardware emulation facilities on x86-32 to support ancient 32-bit CPUs that very very few people are using with modern kernels. This compatibility glue is sometimes even causing problems that people spend time to resolve, which time could be spent on other things." "This will be the first time Linux has dropped support for a major chip family since 2012, when Linux stopped supporting the 386 family," notes ZDNet's Steven Vaughan-Nichols. "Moving forward, the minimum supported x86 CPU will now be the original Pentium (P5) or newer, requiring the presence of the Time Stamp Counter (TSC) and the CMPXCHG8B (CX8) instruction. These features are absent in the older 486 and early 586 processors, such as the IDT WinChip and AMD Elan families." That said, you can continue running Linux on Pentium CPUs, but you'll have to "run museum kernels," as Torvalds pointed out in 2022 when he first floated the idea of ending support for 486.

Delta Air Lines Class Action Cleared For Takeoff Over CrowdStrike Chaos

A federal judge has allowed key parts of a class action lawsuit against Delta Air Lines to proceed, stemming from massive flight disruptions caused by CrowdStrike's faulty Windows update in July 2024. The Register reports: Delta blamed its reliance on Microsoft software and the CrowdStrike incident for its woes. However, according to the plaintiffs in the action (PDF), both companies offered the airline assistance, which Delta turned down. Customers of the Atlanta-based carrier affected by the delays and cancellations claim they struggled to secure refunds and compensation from the airline. The plaintiffs allege that "although Delta offered reimbursement of eligible expenses through their website and app, Delta failed to clarify that the customer would only be receiving a partial reimbursement." "Furthermore, Delta did not disclose to its customers that acceptance of the partial reimbursement would release any legal claims the customer may have against Delta until after the customer 'click[ed] on the button to accept the partial reimbursement.'" The action concerns both US domestic and international travel. The former is covered by US Department of Transportation rules, which require airline agents to "inform customers of their right to a refund ... before making an offer for alternative transportation, travel credits, vouchers, or other compensation in lieu of refunds." The latter claims come under the Montreal Convention, which is designed to be a single, universal treaty to govern airline liability. Delta, which estimated its operational losses at around half a billion dollars due to the outage, sought to dismiss the complaint. While the US District Judge, Mark H. Cohen, granted the airline's motion to dismiss some of the claims, he permitted others to proceed. These were Count I (breach of contract based on failure to refund) and Count XII (violation of the Montreal Convention).