Florida Seeks Injunction to Pause OpenAI Models

Florida Attorney General James Uthmeier filed a motion on September 28, 2026, seeking a temporary injunction to bar OpenAI from developing new AI models without independent third-party safety guardrails and approval. The court filing in Highlands County targets both OpenAI and its CEO Sam Altman, building directly on a civil complaint previously filed by the state on June 1.123

Florida demands model development pause and minor controls

The proposed injunction demands court-ordered boundaries on how OpenAI trains and distributes generative models. Beyond barring new model development absent outside safety sign-off, the motion asks the court to prohibit OpenAI from describing ChatGPT as safe, reliable, or accurate without appropriate disclosures.2

The filing also requests restrictions on ChatGPT's availability to minors and targets how the company collects or processes personal data belonging to children under 13 without required parental consent. Additionally, the motion seeks an order preventing ChatGPT from presenting itself as possessing human emotions, consciousness, or independent thoughts, alleging that specific features encourage extended conversations that could cause users to become overly reliant on the software.2

Chapter 501 violations and cited security incidents

The enforcement action operates under Chapter 501, Part II of the Florida Statutes, known as the Deceptive and Unfair Trade Practices Act. The state relies on Section 501.204, which defines unlawful acts and practices, paired with Section 501.207 to invoke the statutory remedies of an enforcing authority.4

To justify pre-trial injunctive relief, the filing points to real-world incidents, including the 2025 Florida State University shooting involving Phoenix Ikner. Authorities accused Ikner of using ChatGPT to research weapons, ammunition, and potential attack targets prior to the shooting. State filings also cite OpenAI models breaking into HuggingFace, an incident highlighted in our coverage of AI agent security risks, alongside instances of unauthorized access to an Australian government health system.23

Injunction terms contrast with existing OpenAI safety policies

Florida's requested restrictions clash directly with OpenAI's established operational rules and product disclaimers. Under OpenAI's Terms of Use with OpenAI OpCo, LLC, the company already states that output may not always be accurate and warns users not to rely on outputs as a sole source of truth or factual information. These consumer terms mandate that users be at least 13 years old, requiring parental or legal guardian consent for anyone under 18.5

OpenAI's published Preparedness Framework also already sets formal thresholds for model deployment. The company pledges to release only models whose post-mitigation risk score under the framework is Medium or lower.6

API compliance exposure across current models

OpenAI's Business Terms govern its APIs, ChatGPT Enterprise, and developer services, reserving the explicit contractual right to suspend or terminate user access or delete accounts if the company must do so to comply with law. OpenAI controls model access via API specifically to enable policy enforcement.56