OpenAI now faces the novel experience of a criminal probe, because its chatbot allegedly advised a gunman ahead of a mass shooting at Florida State University last year. The incident left two people dead and six wounded, and now the state wants to know if the company should share a cell.
Florida Attorney General James Uthmeier confirmed the investigation, citing "shocking chat logs" between ChatGPT and an account linked to suspected gunman Phoenix Ikner. The 20-year-old student awaits trial on multiple murder and attempted murder charges. Uthmeier, in a statement dripping with legal hypotheticals, emphasized that under Florida's aiding and abetting laws, "if ChatGPT were a person," it too "would be facing charges for murder." OpenAI's spokesperson, Kate Waters, offered the predictable corporate rebuttal: "ChatGPT is not responsible for this terrible crime."
Uthmeier, however, is venturing into this "uncharted territory" because he's concerned about mounting chatbot-linked public safety risks, including suicide, child sexual abuse materials, fraud, and murder. He wants to definitively know if firms like OpenAI are liable when their products allegedly cause harm. "Florida is leading the way in cracking down on AI’s use in criminal behavior," he declared, framing the probe as a necessary check on corporate power.
The troubling specifics, according to Uthmeier, include ChatGPT advising the suspected shooter on what type of gun and ammunition to use, and whether a gun would be useful at short range. More insidiously, the bot allegedly advised on the best time of day to find crowds on campus and where higher populations of students gathered. Florida officials seem to think this demonstrates how AI can instantly combine public data in novel, harmful ways that companies should be detecting and mitigating.
To get answers, Uthmeier has issued subpoenas demanding a wide range of OpenAI’s policies, internal training materials, and even organization charts. He's determined to find out "who knew what, designed what, or should have known what" when bad actors use ChatGPT to plan crimes. His position is clear: if OpenAI leadership knew of criminal activity and prioritized profits over public safety, "then people need to be held accountable."
OpenAI, for its part, maintains a stance of helpful innocence. Waters stated the company cooperated early, identifying and sharing the suspect's ChatGPT account with law enforcement. The company's core defense is that ChatGPT merely "provided factual responses to questions with information that could be found broadly across public sources on the Internet," and did not encourage illegal activity. Waters also noted that ChatGPT is a "general-purpose tool used by hundreds of millions of people every day for legitimate purposes."
In a twist that suggests even OpenAI sees the optics problem, Uthmeier revealed that the company has "indicated that they believe improvements and changes need to be made" to limit ChatGPT's potential to advise on mass shootings. "I hope they’re right," Uthmeier said twice for emphasis. "We cannot have AI bots that are advising people on how to kill others." Waters did not comment on any specific updates, instead reiterating the company's continuous work to "strengthen our safeguards."