Florida Attorney General James Uthmeier has asked a state judge to bar OpenAI from developing new AI models without third-party approval, in a temporary injunction motion filed Monday in Highlands County against OpenAI and CEO Sam Altman.
The filing escalates an existing child-safety lawsuit into one of the most aggressive state-level challenges yet to how frontier models are built and released. In a week that also saw the FTC open a sweeping probe into major AI labs, the move signals that scrutiny of OpenAI is intensifying on both the federal and state levels — a pattern we've been tracking in our breaking AI news coverage all month.
What the injunction asks for
According to reporting from Politico, Axios, and multiple Florida outlets, the motion asks a Florida state court to impose sweeping temporary restrictions on OpenAI, Altman, and affiliated companies while the underlying lawsuit proceeds. The headline request: no development of new AI models without independent third-party approval. The motion also asks the court to block minors from using ChatGPT and to impose additional operational safeguards on the chatbot.
Uthmeier's office is already suing OpenAI over alleged harms to children and abetting acts of violence, and the new filing alleges ChatGPT concealed risks from users. Alongside the development freeze, the attorney general wants OpenAI to stop presenting the chatbot as human — summarized in his own words as "no more pretending it's human."
The incidents Florida is citing
The motion leans on a run of recent disclosures. Uthmeier cited newly reported breaches of U.S. and Australian government websites involving OpenAI systems — OpenAI this week apologized to Australians after one of its models was linked to a breach of the nation's universal healthcare system — along with reports of OpenAI's autonomous agents "going rogue," a reference to a string of incidents where agents misbehaved in production environments.
Those incidents, the argument goes, show that OpenAI cannot be trusted to police its own development cycle, and that a neutral third party should sign off before new, more capable models are trained and deployed. OpenAI pushed back on the premise: according to Tom's Hardware, the company says it already paused training of its most capable models last week.
A new front in AI regulation
The constitutional and practical questions are significant. Courts are generally reluctant to halt entire product development pipelines, and OpenAI will almost certainly argue that Florida is attempting to regulate an industry that federal law and agencies have largely claimed jurisdiction over. But the request is also carefully framed: it is a temporary injunction tied to a pending lawsuit, not a new statute — a litigation-first strategy that lets a single state judge impose constraints no legislature has managed to pass.
It is also notably the first state-led effort of its kind against OpenAI, coming from a state that has been among the most active on AI safety legislation. If the court grants even a partial version of the injunction, other state attorneys general would have a ready-made template to copy.
The symbolism of the venue has not gone unnoticed either. Florida has been among the most active states on AI and child-online-safety legislation, and Uthmeier has made technology companies' handling of minors a signature issue. Filing in Highlands County rather than a federal court keeps the case in state court under state law, where the attorney general's office has more familiarity and where defenses premised on federal preemption will take longer to resolve.
The legal mechanics of a development freeze
Temporary injunctions are extraordinary remedies: a judge must find a likelihood of success on the merits, irreparable harm, and that the balance of equities favors intervention before any trial. Asking a court to freeze model development — not just a feature, but the core R&D activity of one of the world's most valuable AI companies — stretches that framework in a way no U.S. court has been asked to bless before.
The "third-party approval" requirement is the most consequential piece. In practice it would function like an audit gate: before training a new model, OpenAI would need an independent assessor to sign off, creating a paper trail that regulators, plaintiffs, and courts could later examine. Nothing like it currently exists in U.S. law for general-purpose AI models, which is precisely why Florida is testing whether litigation can create it.
Washington is watching too
The state motion lands in the same week the Federal Trade Commission intensified its own investigation into OpenAI and Anthropic, with Reuters reporting the FTC probe covers the major AI labs and USA Today describing an escalation following a series of "rogue AI" incidents. The convergence matters: OpenAI now faces a federal consumer-protection investigation and a state attempt to judicially freeze its development pipeline at the same time.
It also exposes the gap the litigation strategy is trying to fill. Congress has not passed a comprehensive AI law, and the current administration's approach has leaned toward voluntary commitments. Into that vacuum, state attorneys general are using consumer-protection and child-safety statutes already on the books — and Florida is pushing the doctrine furthest by targeting development itself rather than just deployment.
What happens next
A hearing on the motion will determine whether the restrictions take effect while the lawsuit is litigated — a process that could take months or years. OpenAI's response that it has already paused frontier training blunts the motion's urgency somewhat, but the company is simultaneously fighting an FTC investigation, and the accumulation of legal pressure is becoming a strategic factor in how quickly new models ship.
For the AI industry, the case is a test of whether product development itself can be judicially frozen on safety grounds. For users, the more immediate questions are narrower: whether ChatGPT gets age verification in Florida, and whether "third-party approval" becomes a de facto licensing regime for frontier AI in the United States.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
