Florida Attorney General James Uthmeier asked a state court on September 28, 2026, to temporarily restrict OpenAI’s model development and aspects of ChatGPT’s design and availability. The motion advances an existing lawsuit rather than opening a new case, according to reporting by News4JAX and the News Service of Florida. (News4JAX)
The proposals address distinct activities: developing AI models without independent safety safeguards, and operating a chatbot whose presentation allegedly encourages misplaced trust. They also target minors’ access, safety representations and features that prolong conversations. Those are requests for judicial action—not restrictions established by the filing. (WUFT / News Service of Florida)
News4JAX reported the request as pending on September 28 in the Circuit Court of the 10th Judicial Circuit in Highlands County. Secondary accounts conflict: Bloomberg Law places the underlying case in Sarasota County. A copy of the motion retrieved for this revision identifies Highlands County in its caption, supporting News4JAX’s account. (News4JAX)
The current court docket could not be independently checked, and any subsequent orders or filed opposition remain unavailable for this draft. The retrieved motion helps address the venue discrepancy but does not establish the case’s disposition on September 29, 2026. The September 28 account should therefore not be read as confirmation that the request remains pending or that an injunction has been granted.
Development safeguards and chatbot presentation
The development proposal would prevent OpenAI from developing new models without third-party-approved safety guardrails, according to Ars Technica. The exact models covered, approval arrangements and geographic reach remain unverified for this draft. The reported proposal should not be described as an unconditional, permanent ban on AI research. (Ars Technica)
Separately, Uthmeier alleges that first-person language and simulated emotion misleadingly encourage users to treat ChatGPT as a trustworthy friend. The Verge reports that Florida wants to prohibit giving the chatbot false human attributes. That allegation is not a judicial finding that particular wording establishes deception or liability. (The Verge)
The distinction matters for interpreting the request. A restriction on future model development and a restriction on how an existing chatbot communicates would address different conduct. Changing conversational presentation would not, by itself, answer the state’s demand for independent development safeguards; a development restriction would not necessarily resolve its objections to user-facing features. This is an interpretation of the reported requests, not a prediction of an eventual order. (WUFT / News Service of Florida)
Children’s access and OpenAI’s response
News4JAX also describes a proposed prohibition on offering ChatGPT to minors in Florida. A separate request concerns collecting or processing personal information from children under 13 without complying with parental-consent and other requirements under federal children’s privacy law and Florida law. The under-13 threshold should not be confused with the broader proposed access restriction. (News4JAX)
Other requested measures concern representations that ChatGPT is safe, reliable or accurate, warnings about risks, and prompts or features that extend engagement. These remain descriptions of the relief Florida seeks, rather than verified instructions companies must now follow. (News4JAX)
OpenAI spokesperson Drew Pusateri told The Verge that training of its most capable models was paused and would resume only when the company was confident additional safeguards were in place. He supported industry-wide government safety standards. The published response was checked for this revision; no filed opposition was retrieved. (The Verge)
Broader scrutiny of chatbot relationships
Florida’s product-design concerns sit alongside a separate federal examination of AI companions. In September 2025, the Federal Trade Commission announced an inquiry into seven companies, including OpenAI, Alphabet and Meta, seeking information about chatbot effects on children and teenagers. The agency specifically identified humanlike communication and relationships as areas of concern. (Federal Trade Commission)
The inquiry asks how businesses monetize engagement, assess potentially negative effects before and after deployment, disclose risks, and handle information from conversations. Those subjects overlap with questions raised by Florida’s reported requests, but the federal inquiry is a separate proceeding and does not substantiate the state’s allegations. (Federal Trade Commission)
The FTC described its information-gathering authority as permitting broad studies without a specific law-enforcement purpose. That distinction matters: a regulator seeking information is not equivalent to a court finding wrongdoing, just as Florida’s motion is not equivalent to an injunction. Neither development should be reported as proof that all conversational AI products cause the alleged harms. (Federal Trade Commission)
California has pursued legislation addressing related design questions. On October 13, 2025, Governor Gavin Newsom signed a package that included companion-chatbot safeguards. The governor’s announcement described disclosures that interactions are artificially generated, break reminders for minors, and protocols addressing expressions of self-harm. Those measures provide a policy comparison, not a template for what Florida’s court will order. (Office of Governor Gavin Newsom)
For builders and enterprise buyers, the relevant distinction is between development oversight and user-facing safeguards. The reported Florida proposals reach both; the federal inquiry and California legislation also examine how people encounter and understand chatbot products. Their different legal forms—litigation, an agency study and legislation—should remain separate when assessing potential business consequences. (WUFT / News Service of Florida)
What remains unverified
For this case, the reporting and retrieved motion establish that Florida requested restrictions, while published coverage records OpenAI’s public response. They do not establish the current disposition of the motion. The docket, any resulting order and any filed opposition remain necessary to complete the procedural picture and determine whether enforceable restrictions have followed. (Plaintiff’s Motion for Temporary Injunction, via Bloomberg Law)
