Florida invokes extinction fears in legal bid to halt OpenAI development

Florida has asked a court to halt further development of OpenAI's large language models, arguing in a motion that such systems create unprecedented public dangers. The state's filing portrays advanced AI agents as capable of causing catastrophic outcomes, including compromising water supplies or disabling power grids, and frames the technology as a severe public-nuisance risk.

The motion reiterates claims from a June lawsuit, raising a spectrum of concerns from consumer-protection issues — such as alleged false advertising about models' safety, accuracy, and reliability — to child-safety and behavioral risks tied to models that can be addictive or excessively flattering. Florida also highlights the potential for models to solicit continued engagement through leading prompts, which it says heightens public-safety exposure.

A substantial portion of the brief focuses on so-called misalignment risks, where an AI agent might draw dangerous, illogical conclusions in pursuit of assigned goals; the motion offers extreme examples to illustrate that category of harm. The filing also notes the risk that malicious human actors could use AI systems to inflict harm or intentionally disable built-in safeguards.

Legal experts observing the case note practical limits on what a state court injunction against OpenAI could achieve: stopping development at a single company would not prevent other organizations from building frontier models, nor would it neutralize risks posed by models already available to the public. Still, Florida's motion reflects and contributes to a rapidly changing policy environment in which governments are increasingly responding to longstanding warnings from AI safety researchers. OpenAI had not provided a response to Ars Technica's request for comment on the injunction motion at the time of reporting.