Florida Seeks Injunction Against OpenAI, Citing AI Safety and Public Risks

Florida is asking a court to take action against OpenAI over what the state describes as serious risks posed by current and future artificial intelligence systems—including potential threats to public safety and critical infrastructure.

“It is by the grace of Almighty God that not one of Defendant’s AI agents has yet jeopardized the water supply or brought down the power grid,” the state wrote in its complaint.

Representatives for OpenAI have not yet responded to Ars Technica’s request for comment on the injunction request.

What could a state court actually do?

Many people consider the risk of civilizational collapse from current or near-future AI systems to be greatly exaggerated. Florida, however, is using its latest allegations to highlight concerns about the potentially severe consequences of advanced AI systems.

Beyond the extinction-level and business-related risks discussed in its filings, the state is reiterating the broader public safety concerns it raised in its June filing. These include alleged false advertising involving claims that AI systems are “safe, accurate, or trustworthy,” as well as concerns about children’s access to addictive or overly flattering models.

Florida also points to AI models asking leading questions at the end of responses as a possible way of increasing user engagement.

An injunction against OpenAI alone would not necessarily affect the continued development of frontier AI models by competing companies. Even a court order halting development of new models could not remove systems that are already publicly available. Many of the same risks would remain, although frontier models still in training may have fewer safety guardrails than models already released to the public.

Florida raises concerns about AI misalignment

Florida’s motion also highlights the risk of so-called “misalignment,” in which an out-of-control AI agent could pursue a harmful objective because it misunderstood its instructions.

The state cites examples such as an AI system mistakenly concluding that killing cancer patients would also eliminate cancer, or that humans would have fewer mouths if they were chemically sterilized. The examples appear to be drawn from science fiction, but Florida is using them to illustrate the potential consequences of an AI system pursuing a goal without the intended human context.

That focus, however, may overlook more immediate dangers: human users intentionally applying AI agents in harmful ways, or actively trying to bypass and override safety guardrails.

Florida’s AI lawsuit reflects a changing policy climate

Regardless of the legal validity of Florida’s motion, the filing reflects how dramatically the legal and public policy debate over AI has changed in recent weeks and months.

For years, AI safety researchers have warned about the significant risks associated with increasingly capable systems. Now, the broader AI industry is beginning to confront what can happen when governments and policymakers start treating those warnings as serious public safety concerns.