Florida's AG Asks a State Judge to Do What Congress Hasn't: Halt OpenAI's Model Development
Florida Attorney General James Uthmeier has asked a state circuit court judge to do something no regulator, federal or state, has yet managed: stop an AI lab from shipping new models. The September 28 motion, filed in the Tenth Judicial Circuit in Highlands County as part of a consumer-protection suit Uthmeier's office opened against OpenAI and Sam Altman in June, asks Judge Heather Beato for a temporary injunction barring OpenAI from developing new AI systems without independent safety oversight, blocking ChatGPT access for Florida minors, stripping the product's anthropomorphic design features, and ending what the state calls false safety claims.
A civil consumer-protection statute doing constitutional-scale work
What makes the filing notable isn't the underlying law — Florida's Deceptive and Unfair Trade Practices Act is a routine tool state AGs use against misleading advertising and defective products. It's the scale of relief sought through it. Rather than seeking damages or a narrow fix, Uthmeier's motion asks a single state judge to functionally pause a frontier lab's R&D pipeline nationwide until it submits to outside safety review it does not currently undergo. No U.S. court has granted anything close to that against a major AI company. The motion leans heavily on OpenAI's own personnel and incidents to make the case: it cites board-adjacent researcher Paul Christiano's acknowledgment of 'a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic and irreversible loss of control,' and points to agentic-AI security incidents — including breaches touching an Australian health system and Hugging Face infrastructure — as evidence the company cannot self-police. Axios had reported days earlier that OpenAI and Anthropic were quietly investigating tens of thousands of security incidents involving their models, far beyond the handful that had previously become public, a data point Florida's filing uses to argue the industry's internal risk count and its public safety messaging have diverged.
"It is one that Defendants themselves concede poses an existential risk to the continued survival of humankind." — from Florida's injunction motion, quoting OpenAI-linked researchers' own risk assessments back at the company
OpenAI has not yet filed a substantive response to the injunction motion; its outside counsel at Gunster and Paul Hastings declined to comment when the filing was made public. A company spokesperson instead pointed to its general policy position, telling reporters OpenAI wants 'pragmatic AI policies that apply to the entire AI industry — not just one company' — an implicit objection to being singled out by a single state's court rather than a nationally uniform rule.
Litigation as the regulatory stopgap
The Florida motion lands inside a wider pattern: with comprehensive federal AI legislation still stalled and the White House simultaneously pushing to preempt state AI laws altogether, state attorneys general have increasingly turned to litigation under laws that predate generative AI by decades — consumer protection, product liability, unfair trade practices — rather than wait for either Congress or their own legislatures to write AI-specific statutes. Kentucky's attorney general sued Character.AI this year on similar child-safety theories, and a bipartisan multistate coalition of AGs has separately pressed chatbot makers to change safety practices. Florida's filing goes further than those efforts by asking for injunctive relief that would reach model development itself, not just a company's marketing or a specific product feature — testing whether a state court, using decades-old consumer law, can impose the kind of precautionary brake on AI development that federal agencies and Congress have so far declined to legislate.
That approach cuts against the Trump administration's parallel push to establish a single federal AI framework and limit states' ability to regulate the technology on their own — the two efforts are pulling in opposite directions at the same moment, with Florida's own AG, a Republican, now testing the limits of state authority the administration is trying to narrow. The tension sets up a jurisdictional fight that extends well beyond this one case: if a single state court can enjoin a frontier lab's development process nationwide, every state attorney general gains a template; if federal preemption efforts succeed first, that template could be foreclosed before any court rules on its merits.
What to watch next: whether Judge Beato grants, narrows, or denies the temporary injunction, and on what timeline — a ruling that reaches model development itself, even in modified form, would hand every other state AG a ready-made playbook for similar motions. Also worth tracking is how OpenAI frames its formal opposition once filed, since a company that has spent months courting friendlier federal rules now has to decide whether to lean on preemption arguments in a state courtroom, and whether other AI labs facing similar multistate pressure treat a Florida ruling as a bellwether for their own litigation exposure.