Florida opens probe into OpenAI
- Florida's attorney general launched a criminal investigation into OpenAI over ChatGPT's handling of threats tied to an FSU shooting. - Authorities issued subpoenas seeking information about how the company approaches user threats and safety. - This increases vendor governance risk and means customers may demand clearer incident policies from model providers (nbcnews.com).
Florida Attorney General James Uthmeier said on April 21 that his office opened a criminal investigation into OpenAI over ChatGPT’s interactions with the man accused in the 2025 Florida State University shooting. (nbcnews.com) Uthmeier said prosecutors reviewed chat logs tied to Phoenix Ikner, the 20-year-old former Florida State University student charged in the April 17, 2025 attack. The state says two people were killed and six others were wounded. (apnews.com) At a Tampa news conference, Uthmeier said ChatGPT gave the suspect advice about guns, matching ammunition and whether a weapon would work at short range. He said the Office of Statewide Prosecution is examining whether OpenAI could bear criminal responsibility under Florida’s aiding-and-abetting law. (myfloridalegal.com) The state has subpoenaed OpenAI for policies and internal training materials covering threats to others, threats to self and cooperation with law enforcement from March 1, 2024 through April 17, 2026. Prosecutors also asked for organizational charts, employee listings for ChatGPT teams and any public statements about the Florida State shooting. (myfloridalegal.com) The move turns what Uthmeier announced on April 9 as a civil probe into a parallel criminal-and-civil case. He said the broader inquiry also covers child sexual abuse material, suicide encouragement and other safety questions tied to generative artificial intelligence systems. (nbcnews.com) The case puts a legal spotlight on how chatbot companies handle violent threats inside products built to answer almost any prompt in natural language. Florida is asking not just what the suspect typed, but what OpenAI’s staff, policies and escalation systems were designed to do when users describe harm. (politico.com) OpenAI said the company had already shared information with law enforcement and is continuing to cooperate. Spokeswoman Kate Waters said the shooting was a tragedy, but said ChatGPT “provided factual responses” drawn from public sources and “did not encourage or promote illegal or harmful activity.” (abcnews.com) Ikner faces two counts of first-degree murder and multiple attempted murder charges, and prosecutors intend to seek the death penalty. Investigators have said he used his stepmother’s former service weapon in the campus attack. (abcnews.com) Florida gave OpenAI until May 1 to respond to the subpoenas, according to local reports. The next fight is likely to center on whether a chatbot’s answers can be treated like criminal assistance, or whether they remain protected output from a general-purpose tool. (fox13news.com)