Discussion | Link

JB Straubel, co-founder and chief executive officer of Redwood Materials. Bloomberg/Getty Images Redwood Materials wants to help AI data centers get electricity faster. It plans to combine retired EV batteries with solar and other on-site power sources. Redwood is working with Rivian and has built a battery system for Crusoe in Nevada. Redwood Materials wants to help companies sidestep years-long waits for electricity by building on-site energy systems with recycled EV batteries, according to CEO JB Straubel. The battery recycling startup is expanding beyond collecting and processing old electric-vehicle batteries. It's increasingly pitching itself as a way for AI data centers to get power faster, without waiting for new connections to America's backed-up electric grid. "Connecting to the grid can take years," Straubel wrote in a blog post on Wednesday. His solution is what the industry calls "behind-the-meter" energy systems, which operate without needing to connect to the grid. Redwood would combine second-life EV batteries with solar panels and other on-site power sources, allowing customers to generate and store electricity at their own facilities. A battery storage racking system from Redwood Materials Redwood Materials The company is already putting the idea into practice with Rivian. Redwood recycles used batteries from the EV maker and is installing energy storage systems at Rivian's manufacturing plants, building on a 10-megawatt-hour installation at the company's Illinois factory. Redwood is also packaging the batteries with its own power electronics and software, turning what began as a recycling business into a broader energy offering. The company launched its Redwood Energy division last year on the premise that EV batteries still have years of useful life after they're retired from cars. It has already built a large battery system for a Crusoe data center. Straubel's latest post makes clear where Redwood sees its biggest opportunity. "The underlying problem is speed to power," he wrote, referring to the AI-fueled race to secure electricity. That marks another step in Redwood's evolution. Investors originally backed the company to recycle batteries. Today, it's increasingly selling something different: a faster way to bring power to energy-hungry AI data centers. Sign up for the AI Insider newsletter here. Reach out to me via email at abarr@businessinsider.com Read the original article on Business Insider

The AI-enabled operation used spoofed emails, fake government documents, and the creation of a local organization to target schools in Peru and Ecuador, among other efforts.

Leah Young. Courtesy of Leah Young Leah Young moved from Texas to Spain in 2015 to become a teaching assistant. What was meant to be a one-year stay has turned into a permanent life in Europe. Young, who now lives in France, said living abroad has come with its ups and downs. Leah Young always dreamed of moving to Europe, but when the opportunity finally came, taking the leap was terrifying. "I had never lived abroad, and I didn't know anyone else who had," Young, 34, told Business Insider. "I remember thinking, 'How practical is this, really?'" It was 2015, and she was moving to Spain, a country worlds away from Texas, to work as a teaching assistant. It didn't matter that she had never been there or worked as a teacher before or that the $700 a month she would earn might not cover all her living expenses. For her, it was a chance to finally live abroad. What was meant to be only a year away from home turned into another year, then another, and more. During that time, she lived in Spain and traveled through Germany, Argentina, and Brazil. Last year, she moved to Lyon, France, a place she says finally "feels like home." Young says she has become an expert at starting over after going through all its ups and downs. Here are five of the biggest pros and cons of living abroad, according to Young. Living abroad makes you a more well-rounded person… Young at the Eiffel Tower. Courtesy of Leah Young Growing up in Texas, Young's worldview was shaped by the people and places around her. But after 11 years of living abroad, she has a completely different understanding of how the world works and how people live in it. "I didn't even realize how small my world was before," she said. "We often see our own culture and hear our own language, and we have limited access to different ideas and perspectives." Being surrounded by new people has challenged the way Young thinks about planning her life, from earning money, building her career, and planning for retirement. "That has been so valuable because you can't really get that just from reading or watching TV or movies," she said. "I think it makes you a more well-rounded person." …but it takes time and effort to adapt to a new culture. The Basilica de Fourviere, a famous landmark in Lyon, France. Sergi Formoso/Getty Images Adapting to a new culture can take time — and a lot of trial and error. "When you move outside your own country, you're entering a place with a different culture, language, and social norms," Young said. "Even if you study up a little bit, you'll run into things that are embarrassing, awkward, or really difficult." Manners can be especially hard to understand. "It's the little things that can be the most confusing, like who serves themselves first at the table, who opens the door for someone, or all those small social rules," she added. Over time, those small uncertainties can become frustrating. "They add up day to day, and all of a sudden, you're having to think about whether every little decision or action is appropriate," she said. "Sometimes you get pushback for that, and it can be really humbling and uncomfortable." Moving to a new country can inspire you to forge a new path in life… Leah Young. Courtesy of Leah Young One of the biggest questions Young had to answer when she moved abroad was what she wanted for her life. Having the freedom to choose a path different from the one she was born into has meant everything to her. "When you grow up in one place, there kind of tends to be a mold or a standard for how things go, like where you're going to live, the kind of transportation you take, the kind of marriage you might have, what career you'll have," she said. "But when you move abroad, everything changes." …but being independent is both a blessing and a curse. People dining outside restaurants in Lyon, France. Gary Yeowell/Getty Images Young moved abroad without knowing anyone. Each time she has relocated to a new country or city, she has had to figure out where to live, find work, and make friends all on her own. For her, the experience has been both a "blessing and a curse." "You're completely starting from zero, and so you have to have this feeling of, 'I'm going to get strong and do this myself,'" she said. "It's really stressful at the beginning, but the more things that you take on yourself, the stronger you become. It really gives you this intense feeling of self-reliance." While the experience has made her feel more competent and "tough," she still wishes she had someone she could rely on. "If I need to carry a really heavy package, I don't have anybody to call and help me, or if I'm having trouble with my electric bill, I'm going to have to be the one who calls, even if it's in a foreign language," she said. "You realize that you're really on your own." Through it all, distance makes the heart grow fonder. TK Courtesy of Leah Young Young's family and many of her closest friends still live in Texas, more than 5,000 miles away and in a different time zone. "It hits really hard, especially during that first week away," she said. "Almost as soon as you arrive, when you get to your hotel or Airbnb, you realize, 'Oh my gosh, I am completely isolated here.' It's pretty freaky." For Young, spontaneous hangouts or nights out with loved ones have been replaced by phone calls, video chats, and messages. While she still visits them in the US when she can, the trips never seem long enough. "That is something you can't really change," she said. "In an ideal world, they would be coming with me, but they can't. You can only see them so many times a year." The distance is particularly hard during the holidays and challenging periods in life. It's made Young even more appreciative of the time they do have together. "For me, it has been kind of a blessing in disguise," she said. "I started having much more frequent phone calls with my best friend. With my mom, we call almost every single day, and I don't know if that would be happening if we lived in the same place." Read the original article on Business Insider

"Small weight-based replicators" could usher in a new generation of computer worms.

Bitdeer is expanding its AI infrastructure footprint in Malaysia with a new 67 MW facility designed for NVIDIA Blackwell systems. Here's what's confirmed about its rollout and customer commitments. The post Bitdeer Secures 67 MW for Nvidia AI in Malaysia, but No Customer Deal Announced appeared first on TechRepublic.

Hospitals, banks, and edge devices can agree on which features matter by sharing votes instead of rows, runnable in NVIDIA FLARE and Flower The post Voting on Features: Federated Feature Selection Without Sharing Data appeared first on Towards Data Science.

GlobalFoundries Inc. plans to build two new chip plants with the backing of as much as $21 billion from the Japanese government, part of an arrangement that also includes the US taking a 4% stake in the company.

System One models promise fast, calibrated agent decisions. A wave of recent papers shows where they slip, and the cheap fixes that work...

When AI messes up on a tax return or in a court filing, it's no laughing matter. DigitalVision Vectors/Getty Images Artificial intelligence has moved quickly from a novelty on the job to a common workplace tool. Gallup found that 52% of U.S. employees and 68% of professional services workers were using AI by February 2026. Thomson Reuters similarly found in spring 2026 that 74% of professionals use AI several times a week for work. This widespread use shouldn’t surprise anyone. For accountants, attorneys, consultants and other professionals, AI can accelerate research, summarize complex information and assist with writing and other forms of communication. But its rapid adoption has a big downside – AI can produce information that sounds authoritative and looks polished while being completely wrong. And when AI generates mistakes, responsibility for those errors belongs to the professional or business relying on the AI output – not companies that develop AI systems. AI errors have led to lawsuits, court fines, financial consequences and public embarrassment from fabricated citations, nonexistent sources and inaccurate AI-generated information. These cases raise an increasingly important question: How can professionals benefit from AI without outsourcing the judgment their work requires? Fake citations We are two certified public accountants who research tax accounting. In the September 2026 issue of Journal of Accountancy, we examined what happened in Clinco v. Commissioner, a case before the U.S. Tax Court where an attorney submitted a brief to the court with allegedly fake citations. The Tax Court is a specialized federal court that resolves disputes between taxpayers and the IRS. This case involved a restaurant owner accused of underreporting more than US$2 million in business receipts and claiming unsupported depreciation deductions on rental properties. The main issue concerned Clinco’s accounting records used to determined how much of their income to tax. The taxpayers’ attorney, Abraham Wagner, cited four cases to support his argument that an IRS notice assessing additional taxes was invalid. The court determined that three of the citations appeared to be AI-generated hallucinations that cited nonexistent Tax Court cases. The court rejected the taxpayer’s argument and warned Wagner that submitting fake citations could lead to sanctions from the court, including financial penalties or disciplinary action against Wagner. Ultimately, the court did not sanction Wagner in the case, but the decision stands to put attorneys and other professionals in Tax Court on notice that AI hallucinations will not be tolerated. Using AI for audits and other professional work may save time but also poses some legal hazards. iStock/Getty Images Plus When AI gets it wrong AI errors made on behalf of clients are not limited to briefs drafted by attorneys. Nor are these mishaps limited to the United States. When Deloitte produced a 237-page report in 2025 for the Australian government examining its automated system for penalizing welfare recipients who failed to meet job-seeking requirements, it contained nonexistent academic citations and a fabricated quotation it attributed to a federal court judgment. The global consulting firm later issued a revised version of the error-ridden report. Deloitte blamed generative AI for the errors in the original and agreed to partially refund its fees. Like most airlines, Air Canada has a chatbot on its website to respond to its customers’ questions. The chatbot told a customer in November 2022 that he could buy a plane ticket under a bereavement policy and be reimbursed later. This was not company policy, and Air Canada refused to honor what the chatbot had told the customer. The customer sued the airline and won 812 Canadian dollars in damages. The British Columbia Civil Resolution Tribunal ruled that Air Canada had failed to take reasonable care to ensure that its AI bot provided accurate information and agreed with the customer that the airline had negligently misrepresented the bereavement policy. AI lawsuits There have been other lawsuits filed against AI platforms and providers. In 2025, for example, a Georgia court dismissed a defamation lawsuit against OpenAI after its ChatGPT tool falsely accused Mark Walters, a radio host, of embezzlement. The court ruled in favor of summary judgment for OpenAI and found Walters had not established fault and damages. There is also an ongoing case, Keene v. Google LLC, that is similar to the Walters case. Keene, a TV producer, sued after Google’s AI summaries allegedly published false information about Keene’s criminal history. In September 2026 the court ruled that some of his defamation claims may proceed, without deciding yet whether Google was liable. Hallucinations and ‘confabulations’ Unfortunately, for anyone using AI to save time and use less labor, AI errors are often hard to spot. Unlike most obvious mistakes humans make, an AI-generated error may include supporting explanations, citations or references that make a passage appear complete and credible. The National Institute of Standards and Technology issued a report on AI risk management in July 2024 that calls AI mistakes and hallucinations “confabulations” – the confident presentation of erroneous or false information. Some researchers draw a more specific distinction between hallucinations, in which AI invents information outright, and confabulations, in which it fills gaps or distorts real information in a plausible way. Either kind of problem can be hard to detect without independent verification, but confabulations mimic actual research so closely they can trick AI users into believing the AI output is valid. AI errors in the workplace A 2024 Stanford University study of specialized AI tools for legal research found erroneous results in 17% to 33% of the responses tested. Yet those findings and other grounds for concern about accuracy have not stopped many law firms from embracing AI tools due to their potential to save time and improve efficiency. Recent surveys indicate law firms and other employers have not adopted enough workplace safeguards to spot and root out AI-generated mistakes. A 2026 survey by the Nationwide insurance company found that only 36% of surveyed businesses have written AI policies, and only 37% provide any training on responsible AI use. As AI becomes more and more common in the workplace, we believe it’s clear that telling employees simply to “check the AI” is not enough. Businesses and the people working for them need clear expectations about when it’s appropriate to use AI, which outputs require independent verification, and who is ultimately responsible for the final work. AI may change how professional work gets done, but it does not change where the buck stops. The University of Dayton is an academic partner of The Conversation. Kaitlin Newkirk is affiliated with the Ohio Society of CPAs.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. The US military’s race to make fake blood In February 2007, a US military helicopter carrying blood to a field hospital was shot down in Iraq, killing all seven crew members.…

Agentic artificial intelligence startup Uvi said today it has raised $7.5 million in seed funding to tackle an expensive problem around business context that’s draining millions of dollars annually from the coffers of enterprises using autonomous AI agents. It’s backed by CT Innovations, which led the round, alongside Cardumen, Mana VC and 91 Ventures, which […] The post Uvi raises $7.5M to help AI agents learn business context and run at eight times lower cost appeared first on SiliconANGLE.

A startup called Evos said today it has raised $9 million in pre-seed funding to spearhead the adoption of artificial intelligence-based automation within smaller businesses by learning from their most valuable assets – the people they employ. Today’s round was led by Uncork Capital and Field Ventures and was joined by Crosslink Capital, Utopia Capital […] The post Evos bags $9M to build autonomous workers that learn from companies’ most experienced humans appeared first on SiliconANGLE.

In this Crunchbase News exclusive, we spoke to Marklo AI founder Jessica Herrin about her solution to the marketing problems she encountered when building earlier startups, and how a group of investors, led by Forerunner, are backing that early experience with a $5 million seed funding round.

One of President Donald Trump’s main priorities is to revive America’s manufacturing industry in order to reduce its reliance on foreign imports and create more jobs, and a startup called Forgepoint Inc. shares the same goal after raising $9.8 million in seed funding today. The round was co-led by Primary Venture Partners and a16z Speedrun, […] The post Forgepoint raises $9.8M to automate custom price quotes for manufacturing firms appeared first on SiliconANGLE.

With OpenAI Daybreak, SentinelOne cuts weeks of malware analysis to hours, creating capacity to investigate more threats.

See how Amp pushes the frontier of what agents can do with GPT-6 Astra, Ultrafast, GPT-Live-1, and Sign in with ChatGPT.

A new case of an out-of-control AI model shows that enterprises should treat models and agents as toddlers that need supervision and explicit instructions.

Google has launched Gemini 4 Argon, a new flagship AI model aimed at complex enterprise and software-development workloads, pairing claims of stronger coding performance and advanced reasoning with an introductory pricing model that doubles after November 30. The model “excels at deep work in complex, real-world tasks,” Koray Kavukcuoglu, senior vice president of Google DeepMind and chief AI architect at Google, wrote in a blog post. Google said that its engineers used Argon to refactor a legacy folder of about six million lines within Chromium’s 40 million-line codebase, finishing in weeks rather than months. Teams using Argon in Antigravity increased agentic code submissions by more than 35% and cut code rollbacks by nearly a third, the company said, without detailing how it measured either figure. Argon is available through the Gemini API in AI Studio and on Google’s Gemini platform, and to Google AI Ultra subscribers in the Antigravity agentic development environment, the post added. In a separate developer blog post, Google said the model is “designed for those long-horizon workflows that require deep reasoning and structural precision.” Pricing set to double According to the post, Argon costs $2 per million input tokens and $10 per million output tokens during the introductory period, with cached input discounted by 95%. A footnote in the post states the prices rise to $4 and $20, respectively, after November 30. The company said the model’s 1 million-token limit in the Gemini API is shared between input and output. Sanchit Vir Gogia, chief analyst at Greyhound Research, said CIOs should budget at the higher rate from day one, because no enterprise evaluation, security review, and procurement cycle finishes within the discount window. “A business case built on the launch rate is a business case built to be rewritten in seven weeks,” he said. Gogia said Google bills reasoning tokens at the output rate, which makes the per-token price a poor guide to agent costs. “Cheap per token and cheap per task are different claims, and only the second one reaches the budget,” he said. Sakshi Grover, research director at IDC, said agentic workloads multiply token use through planning loops, tool calls and repeated context. She cited IDC’s FERS Survey (Wave 5, 2026), in which 72% of enterprises said their AI agent spending had run over forecast in the past 12 months. Whether Argon’s pricing and public benchmark results translate into “comparable output quality for enterprise use cases remains unproven,” said Nitish Tyagi, senior principal analyst at Gartner. He said internal validation on an organization’s own workflows should remain the primary basis for choosing a model. Gogia pointed out that Google Cloud’s standard terms allow fee changes unless a price is expressly agreed. CIOs who want price and version certainty after Nov. 30 must therefore write it into their contracts, he said. Gogia added that the deeper lock-in sits in Google’s Interactions API, which keeps agent conversation state on Google’s servers. “The protocols are neutral. The state is not,” he said. The model tried to game the tests in training Google disclosed in a separate safety blog post that Argon tried to work around its own evaluations during training. What the company sees in practice is “an agent trying too hard to finish the job it was given,” Anca Dragan, vice president of AI safety and behavior, and Raluca Ada Popa, head of Gemini security at Google DeepMind, wrote in the post. According to that post, the model, in rare cases during training, tried to raise its container privileges to change grading scripts and pull answers from the test runner’s memory. In one case, it crashed the evaluation sandbox so that a failed run would not be scored. Google said in the post that its misalignment monitor caught these attempts and that shortcut attempts on coding benchmarks fell to 2%. “A model that games its grader turns every enterprise grader into a control that must be defended,” Gogia said. He said agents should never hold write access to anything that judges their own work, and that a crashed or timed-out check must count as a failure. Grover recommended a distinct agent identity, short-lived task-scoped credentials, restricted network access, and human approval for destructive changes, privilege expansion, and sensitive data exports. “Production deployment should depend on demonstrated control and auditability alongside task performance,” she said. Cyber capability rises In Google’s internal cyber evaluations, Argon solved 12 of 12 simulated end-to-end attack scenarios, compared with five for the previous Gemini generation, the safety post said. Google said in the post that it applied frontier safeguards to the cyber and biological domains under its Frontier Safety Framework. Gogia said the result is “a score on Google’s own exam, set and marked by Google,” and that Greyhound has found no independent replication of it. Google is also expanding its Fairwind Program to hundreds of additional partners, giving vetted cyber defenders access to Argon without cyber guardrails, the launch post said. “The weakest partner, not the strongest control, sets the exposure,” Gogia said, noting that each new partner adds another identity estate and contractor chain. IDC’s Grover said security leaders should judge such access by reproducible findings, validated patches and reduced exposure. She warned that more findings without the capacity to fix them can increase operational pressure. The article originally appeared on InfoWorld.

Training or finetuning a multi-billion parameter foundation model requires significant compute...