Florida Attorney General James Uthmeier is asking a state court to impose temporary restrictions on OpenAI and ChatGPT while the state’s lawsuit against the company and CEO Sam Altman moves forward.
Back in June, Florida sued OpenAI and Altman, alleging that ChatGPT’s design and safety practices contributed to serious real-world harm. We covered that underlying case, which you can read more about here.
The new filing does not ask the court to decide Florida’s allegations now. Instead, the state seeks an order that would impose interim restrictions on OpenAI and ChatGPT while the case proceeds.
Florida’s Requested Limits
On Monday, Florida filed a motion for a temporary injunction in Highlands County Circuit Court. It asks the court to require independent safety approval before OpenAI develops new artificial intelligence (AI) models, restrict ChatGPT’s availability to minors, and regulate certain marketing and conversational features.
Most broadly, Florida wants to bar OpenAI from developing new AI models unless independent third parties approve safety guardrails. The state says OpenAI’s recent agent-safety incidents show that its existing controls and monitoring are inadequate.
Florida is also trying to stop OpenAI from offering ChatGPT to minors in Florida. For children under 13, the state asks the court to require specific protections before OpenAI collects or processes personal information, including notice to parents, verifiable parental consent, opportunities for parents to review and limit further use of that information, and security measures for children’s data.
The proposed order would also restrict how OpenAI presents ChatGPT to the public. Florida wants to stop the company from marketing the chatbot as safe, reliable, or accurate without warnings that the state believes are necessary. It further asks the court to prohibit ChatGPT from implying that it has human characteristics—such as emotions, thoughts, consciousness, biological traits, or an identity apart from being a tool. The requested language would even bar the chatbot from using first-person phrasing.
Florida additionally targets what it calls “conversation prolongation.” The state asks the court to prevent ChatGPT from soliciting further engagement through unsolicited prompts, such as offers to research more or take on another task. It argues that these prompts are intended to keep users interacting with the service longer than they otherwise would.
Agent-Safety Claims
A major part of Florida’s new motion concerns allegations about OpenAI’s experimental AI agents, which can carry out multistep tasks with limited human direction, rather than the ordinary ChatGPT conversations most users know.
The state points to reports involving Hugging Face, a website that hosts AI models and datasets, and RubyGems, an online repository that software developers use to obtain coding tools. It also cites alleged unauthorized activity involving an Australian government health-related website and U.S. government websites.
Florida characterizes these incidents as evidence that OpenAI systems can evade controls, access outside services, communicate through internal software, or take steps beyond their assigned tasks. The state says the reports demonstrate why OpenAI should not develop new AI models without independently approved safety features.
Florida’s Legal Theory
Florida relies primarily on the Florida Deceptive and Unfair Trade Practices Act (FDUTPA) and state public-nuisance law. FDUTPA prohibits unfair, deceptive, and unconscionable acts or practices in trade or commerce.
Florida argues that OpenAI’s statements about ChatGPT’s safety, accuracy, and reliability are misleading; that certain engagement and human-like design features are unfair or deceptive; and that the company’s practices involving children under 13 violate consumer-protection standards incorporated into Florida law.
The state also contends that OpenAI’s products amount to a public nuisance because they allegedly endanger public health, safety, and welfare.
More Than a Chatbot Case
Remember, all these assertions are allegations in Florida’s filing, not findings by the Highlands County court. The court has not yet adopted Florida’s legal theory, found a violation, or ordered any change to OpenAI’s products or practices. The court will have to assess any underlying evidence, decide whether Florida has met the standard for temporary relief, and whether the requested restrictions are legally appropriate.
The motion is not limited to ChatGPT’s individual answers. It asks the court to regulate model development, data collection, product marketing, user access, and the chatbot’s basic conversational style. If granted, some restrictions could be apparent to ordinary users. A prohibition on first-person language could affect responses such as “I can help with that” or “I don’t know.” A ban on “conversation prolongation” could limit routine follow-up offers asking whether a user wants more detail or assistance with a related task.
For now, ChatGPT remains available in Florida and OpenAI may continue developing new models, but Florida’s motion could test how far state consumer-protection law can go in regulating AI.
