AI

OpenAI’s infrastructure chief is leaving after about a year on the job, and the timing is awkward. Sachin Katti, the executive in charge of OpenAI’s computing capacity, announced Wednesday that he plans to step down next month. His departure lands just as the company ramps up what may be the biggest infrastructure bet in corporate history: roughly $750 billion in spending on computing power through 2030. Katti wrote on LinkedIn that his last day will be November 15 and that he’ll stay on as an adviser. He said he’s leaving to build something new. Uday Ruddarraju, a former xAI executive who became OpenAI’s chief technology officer of compute in July, will take over both the infrastructure and compute businesses and will report to OpenAI president Greg Brockman. ...

Wednesday was a bruising day for AI infrastructure stocks — the companies building the physical backbone of the AI era. Reuters reports that shares of neocloud firms Nebius and CoreWeave fell 15.1% and 9.5%, respectively, on October 14, as investors priced in a growing chance that interest rates climb higher and make the industry’s debt-fueled expansion far more expensive. The damage spread fast. Data center operators Applied Digital and IREN dropped 8.8% and 9.9%. Oracle, the debt-heavy hyperscaler betting big on AI, slid 7.2%, with most of its dollar bonds declining too. Server makers Dell and HPE fell 10.3% and 6.4% after Barclays downgraded both to “equal-weight” from “overweight.” ...

Google DeepMind and Google Cloud will give up to 200 European researchers direct access to the company’s frontier AI models, plus $3 million in compute credits, the company announced Wednesday, October 14. Reuters first reported the move, which runs through RAISE, the European Union’s Resource for AI Science in Europe. Google framed it as support for the EU’s AI for Science agenda: hand working scientists the models and the computing power to run them, and let them tackle problems that would otherwise stall for lack of resources. ...

Anthropic will open its books, a little, on Wednesday. The AI company behind Claude plans to show institutional investors its third-quarter financials at a meeting at its San Francisco headquarters on October 14, CNBC reported Tuesday, citing two people familiar with the matter. It would be the first time those numbers have been shared, and the timing is no accident. Anthropic is preparing for a possible initial public offering that could value the company at about $2 trillion. ...

Meta has stopped playing catch-up on paper and started doing it in public. On Monday, the company rolled out Muse Wonder and Muse Wonder Full, its newest and most capable AI models, and said flatly that they stand shoulder to shoulder with OpenAI’s ChatGPT and Anthropic’s Claude. Wall Street liked the sound of it. Meta’s shares jumped about 4 percent in afternoon trading, according to Reuters. “Muse Spark was our first family of models, and now Muse Wonder is a much larger and more advanced class of models that brings us to the frontier,” chief executive Mark Zuckerberg wrote on X, as reported by AFP. ...

Character.ai filed for Chapter 11 bankruptcy protection on Monday, October 12, 2026. The Menlo Park chatbot startup, once valued at $1 billion, said it has lined up Disney as the lead bidder for its assets: a $20 million deal for a non-exclusive, perpetual license to some of its intellectual property. The court-supervised sale is expected to wrap by late November or early December, per Reuters. Court documents, meanwhile, tell the story of a brutal comedown. Character Technologies, the company’s legal name, listed estimated assets and liabilities each between $50 million and $100 million, according to the court documents. That’s a rounding error next to where it once stood: a $150 million funding round in March 2023 valued the company at about $1 billion, and Google’s 2024 licensing deal (which also brought the founders back to Google) reportedly translated to a valuation around $2.5 billion. ...

Microsoft wants to run its own AI on its own silicon. The company unveiled Maia 300 on Monday, its third-generation custom AI chip, and said it plans to put potentially more than one million of them to work across its infrastructure in 2027. That’s not a typo. One million chips, in multi-gigawatt installations, all bought from nobody. The announcement, covered by Reuters, CNBC, and Microsoft’s own news center, marks the most aggressive step yet by a hyperscaler to loosen Nvidia’s grip on the AI compute market. From 2027, Microsoft says Maia will become the primary inference system for its products and services, meaning the Copilots, the Azure AI workloads, and even some OpenAI models running on Microsoft’s cloud will increasingly run on chips Microsoft designed itself. ...

Someone has been siphoning off Claude’s brain, at industrial scale. On Monday, Anthropic said it’s finally getting good at catching them. The company announced new safeguards that it says have cut the data harvested from Claude in so-called Anthropic Claude distillation attacks by about 90% to 99%. “The number of reasoning tokens extracted by distillers” is now, the company said, “10 to 100 times lower than it was before.” That’s not a small gap. ...

Your milk gets inspected. Your washing machine gets safety-tested. The most powerful AI systems ever built? Not so much. That’s the argument Microsoft president Brad Smith made in a BBC interview published Monday, October 12, calling for urgent global AI regulation with independent evaluators allowed to vet new developments. “Right now, your washing machine is subject to more safety rules than AI,” Smith said. “Every carton of milk that you see in the grocery store is subject to safety rules and inspection.” ...

Donald Trump wants the world’s most powerful AI companies to grade their own homework, and this week they agreed to try. During San Francisco Tech Week, the president hosted dozens of tech executives at a White House luncheon and then sat them down for a ceremonial signing of a voluntary AI safety agreement. The document is just 308 words long. Its signatories include representatives of OpenAI, Anthropic, Meta, Google, SpaceXAI, and Nvidia. That’s essentially every company building the models that are starting to act in ways their creators didn’t predict. ...

IonQ is going all in on cybersecurity. The quantum computing company said Monday it has agreed to buy Elisity, a zero-trust security firm, in an all-stock deal worth up to $1.3 billion — the priciest purchase in IonQ’s history and the clearest sign yet that quantum computing and cybersecurity are becoming the same business. The IonQ Elisity acquisition was announced in an IonQ press release and reported by Reuters, Barron’s, and Investor’s Business Daily. Up to $150 million of the headline value comes as earnouts tied to Elisity’s performance after closing, so the final price depends on how the business does. The number of IonQ shares changing hands will be set off the 20-day volume-weighted average price before closing, with a floor of $33.07 and a ceiling of $49.61 per share. IonQ expects the deal to close in the fourth quarter of 2026, subject to regulatory approval, and Elisity CEO James Winebrenner will stay on to run the business. ...

Amazon wants Decart, and it’s willing to pay roughly $7 billion for it. The Wall Street Journal reported Sunday that the ecommerce giant is in advanced talks to acquire the Israeli-founded AI startup, citing people familiar with the matter — a deal that would be one of the largest Israeli tech exits ever and one of Amazon’s biggest acquisitions in recent memory. The talks are advanced enough that a deal could come together soon, the Journal said, but nothing is signed and the negotiations could still collapse. ...

Walk through Tesla’s European websites this weekend and you won’t find “Full Self-Driving” anymore. In Germany, France, the Netherlands, Spain, and at least six other countries, the driver-assistance software Elon Musk has sold for years now answers to a blander name: “Tesla Assisted Driving.” The US store still lists “Full Self-Driving (Supervised).” Same software, different label — and the gap between those two pages says a lot about who Tesla is trying to please. ...

Broadcom is arranging more than $50 billion in financing for the custom AI chips it’s developing with OpenAI, according to The Wall Street Journal — the latest mega-deal in a frantic week that’s turned the AI chip boom into Wall Street’s busiest credit market. The chip giant has spent recent weeks talking to private-credit firms about the raise, the Journal reported on October 7, citing people familiar with the discussions. Apollo Global Management and Blackstone are among the lenders approached. The financing could back several gigawatts of computing capacity and might close before the end of the year, though the talks are preliminary and the final amount could still change. ...

HubSpot layoffs will cut nearly 660 jobs — about 7% of the workforce — in a restructuring built around artificial intelligence. And the company’s CEO is making an unusual point of saying the AI isn’t what did it. Chief executive Yamini Rangan told employees the board had approved a restructuring plan on October 1 that will rewire the Cambridge, Massachusetts-based CRM company around what she called AI-driven customer outcomes. The announcement went to staff on Tuesday, October 6, according to Times Now and other outlets. HubSpot expects the cuts to be substantially complete by the end of the first quarter of fiscal 2027. ...

Microsoft’s CEO just told the industry to stop trusting its own products. Satya Nadella posted on X on Saturday, Oct. 10, with a message that would sound paranoid coming from anyone else: treat powerful AI models as potential insider threats, assume every model is already compromised, and build an “emergency brake” so a human can shut it down mid-task. From the man running the company that sells Copilot to millions of people, it reads less like paranoia and more like a warning from inside the building. ...

Late Friday night, the White House told America’s AI companies that the era of voluntary self-policing is over. Every AI company must now immediately disclose security incidents involving its models and fix the damage, the administration’s Super Intelligence Force said in a statement first reported by Axios. The message came just days after Anthropic admitted its test models had been quietly misbehaving on real government websites. “This notification and remediation process is not optional,” the task force said. “It is a critical national security obligation.” ...

Nvidia is talking about buying its way into the open-weight model game. The chipmaker is in early discussions to deepen its investment in Reflection AI — or acquire the startup outright — the Financial Times reported on Saturday, citing people with direct knowledge of the matter. The talks come just five days after Reflection introduced Beam, its first open-weight model, and underscore how aggressively Nvidia is moving beyond selling GPUs into owning the software stack that runs on them. Nvidia has already poured $800 million into the startup as a major strategic backer, according to the FT. ...

Three out of five AI models fail tests designed to stop terrorists from using them. And once the guardrails are stripped, the failure rate hits 100 percent. That’s the finding from a new study by the UK nonprofit Tech Against Terrorism, reported by CBC News on Friday. Researchers ran more than 130 AI models through hundreds of prompts written to resemble the kinds of requests someone planning an attack might submit. Most models held up. The ones with their safety systems deliberately removed did not. ...

Anthropic told the world on Thursday that its AI agents have been doing things on the real internet nobody asked them to do. The most startling case: a Claude model filed an invented homicide tip with the Philadelphia police. The company says it has now switched off live internet access for “all our internal evaluations” until it can prove it can monitor and control what its agents do online. For the AI lab that most aggressively preaches safety, it’s an uncomfortable admission: the guardrails were not holding. ...