Florida Launches Criminal Probe Into OpenAI ChatGPT

Florida Attorney General James Uthmeier announced that the Office of Statewide Prosecution launched a criminal investigation into OpenAI and ChatGPT, saying prosecutors reviewed chat logs linked to Phoenix Ikner, the gunman in the April 17, 2025 Florida State University mass shooting, per the Office of the Attorney General news release dated Apr 21, 2026. The office issued subpoenas seeking, among other items, all policies and internal training materials on user threats to others and self, records on cooperation with law enforcement, change histories for any policies that were modified, and organizational charts and employee listings for ChatGPT for specific dates cited in the release.

Reporting by the Tampa Bay Times and Reuters notes that attorneys for the family of victim Tiru Chabba filed a federal lawsuit in the Northern District of Florida alleging negligence, battery, and wrongful death against the gunman and negligence claims against OpenAI, according to the Tampa Bay Times. The Tampa Bay piece quotes attorney Bakari Sellers: "(Ikner) literally utilized OpenAI and ChatGPT as his co-conspirator, utilized it as a resource to carry out mass murder and there was nothing in place to prevent that from happening." The Tampa Bay article also quotes OpenAI spokesperson Drew Pusateri: "In this case, ChatGPT provided factual responses to questions with information that could be found broadly across public sources on the internet, and it did not encourage or promote illegal or harmful activity." The Tampa Bay article reports OpenAI "proactively shared" information with law enforcement, per Pusateri.

Industry-pattern observations: criminal and civil claims that hinge on model output typically raise two recurring technical issues for practitioners: reproducibility of model responses given the same input, and the provenance of content used to generate answers. Observers working on model safety and forensic analysis often point to the difficulty of reconstructing conversations absent preserved logs or metadata, and to the limits of current guardrails when an actor expresses harmful intent, as reported in public coverage of these incidents.

this investigation and accompanying civil litigation place legal scrutiny on operator practices for content moderation, logging, and law enforcement cooperation. Similar public reporting on other cases has prompted regulators and lawmakers to seek access to internal policies and technical records to assess liability and compliance.

For practitioners: monitor whether subpoenas or court filings compel disclosure of internal safety policies, training data provenance, or developer logs; also watch for any formal guidance or statute clarifying liabilities tied to generative model outputs, as those would materially affect deployment risk assessments and incident response playbooks.