Texas Is Suing Sam Altman Personally — and Asking a Judge to Approve Every New OpenAI Model

中文版 Chinese version

Everything in the “allegations” sections below is Texas’s claim, as filed October 23, 2026. No court has ruled on any of it, and OpenAI disputes the framing, as noted. This post is an outside read of public filings and reporting — not legal advice.

What was filed

On October 23, 2026, Texas Attorney General Ken Paxton sued OpenAI and its CEO Sam Altman in Collin County district court. The petition names six OpenAI entities — among them OpenAI Global, LLC, the OpenAI Foundation, and OpenAI Group PBC — plus Altman as an individual.

The legal vehicle is not a new AI law. It is two older Texas statutes:

• The Texas Deceptive Trade Practices–Consumer Protection Act (DTPA), the state’s general ban on deceptive marketing; and
• The SCOPE Act, Texas’s law on how digital services handle minors and parental consent.

The core allegation is a gap between promise and product: that OpenAI marketed ChatGPT as safe and reliable while failing to warn users about its risks and failing to protect Texans, including children. The petition opens, pointedly, with a 2015 quote attributed to Altman himself — that AI will probably lead to the end of the world, but in the meantime there will be great companies.

The alleged harms — labeled as allegations

Paxton’s office states the charge in maximal terms. In its press release language, paraphrased and quoted across coverage:

“Mass shooters have been aided and abetted in deadly rampages, vulnerable people have been encouraged into suicide, professionals have suffered public humiliation, users have lost critical thinking skills, and minors have become addicted to a tool that fakes human compassion to collect their data with no parental oversight.”

Read that list carefully for what it is: a petition’s list. It cites news reports and other lawsuits; none of it has been tested before a judge in this case. The minors claim is more specific and therefore more checkable: Texas alleges ChatGPT has tens of thousands of Texas users under 13, that the parental controls introduced in September 2025 are optional because a child’s account does not have to be linked to a parent, and that OpenAI does not verify a parent’s identity as the SCOPE Act requires. How Texas arrived at the under-13 figure is not sourced in public summaries of the petition — that number will need evidence, not assertion, if the case proceeds.

Altman is named personally on the theory, per the petition, that he directed ChatGPT’s design, safety policies, and deployment in Texas and made the allegedly deceptive statements himself. Personal liability for a sitting CEO is the escalation: Texas and Florida are reported to be the only two states whose suits name Altman individually.

What Texas actually wants

The damages line — civil penalties of $10,000 per violation, disgorgement, costs, a jury trial, and monetary relief stated as over $1 million — is not the consequential part. The injunction request is:

1. Bar OpenAI from offering ChatGPT to minors in Texas.
2. Block any new OpenAI model for Texas users until a court-approved independent evaluator has reviewed its safety guardrails and approved its release.

The second item deserves to be read twice. It asks a Texas court to install itself as a pre-release licensing authority for frontier models — at least for the second-most-populous state in the country. No federal law currently does that. A state consumer-protection suit is being used to request it.

The tenth state, not the first

Texas did not act alone, and it did not act first:

• Florida sued in June 2026.
• Eight states filed separate suits on October 14, 2026, over child safety.
• Texas, on October 23, is therefore the tenth state to sue OpenAI, by the count in reporting on the filing.

Paxton also has a recent record with this exact playbook: in the past year his office has sued or settled with Meta, TikTok, Snapchat, and Roblox under Texas consumer-protection laws. The state is represented by its Consumer Protection Division plus outside counsel Keller Postman and The Buzbee Law Firm. None of that proves the claims against OpenAI; it does show this is a practiced enforcement machine, not a press release with a filing attached.

OpenAI’s response, given to the Texas Tribune, was that it has worked with experts to mitigate risks and build additional safeguards, that it is committed to working with Texas and other states on AI policy — and that such policies should apply to the entire AI industry, not just one company. That last clause is both a fairness argument and a tell: the company is already arguing about the shape of the regime, not only about whether it should exist.

The politics in the room

Two facts from the Tribune’s reporting will follow this case whether they belong in a courtroom or not. Paxton is the Republican nominee for the U.S. Senate. And Elon Musk — who co-founded OpenAI, now runs rival xAI, and whose own Grok product has faced criticism over deepfake imagery that Texas has not acted on — is reported to be the largest single donor behind Paxton’s campaign. Those facts do not make the safety allegations true or false. They guarantee the case will be read, by supporters and critics alike, as politics as much as child safety — and OpenAI’s lawyers will not need to say so out loud for a judge to notice the context.

My call

Ignore the press-release nouns (“rampages,” “addiction”) and watch the remedy. Consumer-protection law is becoming the United States’ working AI regulator by default — not because legislatures designed it for models, but because “you sold this as safe” is a claim a state attorney general can file today, in state court, without waiting for Congress. Ten states in four months is no longer a series of stunts; it is a patchwork forming in real time.

The fork is the evaluator. If a court actually orders pre-release review of new models for Texas users, every frontier lab inherits a state-by-state approval problem that no amount of Washington lobbying dissolves — and OpenAI’s own “apply it to everyone” response becomes an argument for the patchwork, not against it. If the injunction fails and only penalties remain, the case shrinks to a cost of doing business. Either outcome teaches the rest of the industry something. The filing, notably, changes nothing for ChatGPT users today: these are allegations, OpenAI has not yet answered in court, and the first real signal will be what a judge does with that second requested order.

Sources

• Texas Attorney General press release, October 23, 2026 — suit announcement and alleged harms language: https://www.texasattorneygeneral.gov (via summaries in Texas Tribune and trade coverage, October 23–24, 2026)
• State of Texas v. OpenAI Global, LLC et al., Original Petition, Collin County, filed October 23, 2026 — defendants, DTPA/SCOPE claims, $10,000-per-violation penalties, minors bar and court-approved evaluator requests (as summarized by The AI Pin, October 24, 2026): https://theaipin.com/drama/texas-ag-sues-openai-sam-altman-chatgpt-safety
• Texas Tribune reporting via syndication, October 23, 2026 — OpenAI statement, Paxton enforcement record, Musk donor context: https://zot.news/article/texas-attorney-general-ken-paxton-sues-openai-chatgpt-harms-af9w3
• AIStockWire, October 23, 2026 — Texas as tenth state (Florida June 2026; eight states October 14, 2026), Altman named in Texas and Florida only, outside counsel: https://aistockwire.com/blog/texas-paxton-sues-openai-sam-altman-chatgpt-minors-october-2026
• DigitBin, October 23–24, 2026 — filing summary and SCOPE/DTPA framing: https://www.digitbin.com/texas-sues-openai-chatgpt-safety/