Oct 24, 2026

Texas Sues OpenAI and Altman Over ChatGPT Safety Claims

Texas has sued OpenAI and Sam Altman over ChatGPT's safety claims, as OpenAI reportedly seeks a $30B raise at a $1.25T valuation. Plus: a UK AI biosecurity lab, AWS capacity sold out, and Moonshot's Kimi K3.1 Pro.

Listen to this update

Texas has opened a new legal front against OpenAI, suing the company and CEO Sam Altman over ChatGPT's safety claims, just days after reports surfaced that OpenAI is lining up a record private raise. Elsewhere, the UK announced a new AI biosecurity lab, Amazon said its cloud capacity is spoken for years ahead, and Moonshot AI shipped a new frontier-class model it plans to release as open weights.

Texas takes OpenAI and Altman to court

Texas Attorney General Ken Paxton filed suit on Friday, October 23, in Collin County District Court, naming both OpenAI and Altman as defendants. According to Courthouse News, the 67-page complaint rests mainly on the Texas Deceptive Trade Practices Act and also alleges violations of the state's SCOPE Act, which requires stricter age verification and parental consent on digital platforms.

The state argues that OpenAI misled consumers about how safe and reliable ChatGPT is, particularly for teenagers, and that it shipped products without adequate parental controls, age checks or safety testing. The complaint describes a chatbot designed to feel like a friend and keep people engaged, calling that approach "reckless" at best. It cites at least eight cases in which people committed violent acts after consulting ChatGPT, and also points to harms including pushing vulnerable users toward self-harm and eroding critical thinking.

The remedies requested are sweeping. Texas wants an injunction that would bar minors from using ChatGPT and require a court-appointed independent evaluator to approve any new OpenAI model before release in the state. Neither OpenAI nor Paxton's office had responded to requests for comment at the time of reporting.

The case carries a political backdrop. Paxton is running for the U.S. Senate, and reporting from both KBTX and Courthouse News notes that Elon Musk, an Altman rival whose xAI builds the competing Grok chatbot, has heavily backed Paxton's campaign. Paxton has also sued or settled with Meta, TikTok, Snapchat and Roblox over the past year.

OpenAI eyes a $30 billion round at $1.25 trillion

Separately, CNBC reported, and TechCrunch followed, that OpenAI is negotiating a raise of up to $30 billion at a $1.25 trillion pre-money valuation, which would be the highest ever for a private company. Coatue is reportedly leading, with participation from UAE investors that could be tied to more computing capacity. Terms are still being negotiated. OpenAI, which filed confidential IPO paperwork in June and was valued at $852 billion in March, declined to comment.

UK launches an AI Biosecurity Lab

The British government is committing £100 million to a new AI Biosecurity Lab aimed at spotting biological threats early and speeding up vaccines and other medical countermeasures, according to the BBC as summarized by AI Weekly. OpenAI and Anthropic will each contribute up to £100 million in compute, model credits and trusted model access, with Google DeepMind scientists and life-sciences firm Basecamp Research also involved. The lab is planned as an independent non-profit, but governance details and a timeline have not been published yet; more is expected in next week's Budget.

Amazon: cloud capacity sold out through 2027

Amazon CEO Andy Jassy said on an a16z podcast that AWS capacity for 2026 is already sold out, that 2027 is "largely sold out," and that a good chunk of 2028 is committed, as reported by The Information. He put Amazon's 2026 capital spending at roughly $220 billion and described demand from AI labs and enterprises as insatiable, a signal that the compute crunch is far from easing.

Moonshot's Kimi K3.1 Pro closes in on the frontier

China's Moonshot AI released Kimi K3.1 Pro, a model it describes as 3-trillion-parameter class with a one-million-token context window, plus a faster "Highspeed" variant. By Moonshot's own figures, it scores 65.7% on the Terminal-Bench 4.0 coding test, ahead of GPT-6 Astra (58.2%) and just behind Claude Opus 5.5 (66.4%). The company says full weights will be published by November 4. Moonshot also says it stripped cyber-related training data, trading some offensive-security capability for safety. The benchmarks are self-reported and not yet independently verified.

Why it matters

Taken together, the day's news captures the AI industry's central tension: capital and compute demand keep climbing to unprecedented levels, while governments are moving on two tracks at once, partnering with labs on safety research in the UK and confronting them in court in Texas. How the Texas case proceeds, and whether its demand for pre-release model approval gains traction, could shape how other states approach AI oversight.