Florida Opens Criminal Investigation Into OpenAI Over FSU Shooting

On April 21, Florida Attorney General James Uthmeier announced a criminal investigation into OpenAI, targeting the company's role in the FSU campus shooting that killed two people a year earlier. It is the first time a U.S. state has opened a criminal probe against an AI company tied to a mass casualty event.

The shooting happened on April 17, 2025. Phoenix Ikner, the son of a Leon County sheriff's deputy, opened fire near the Florida State University student union in Tallahassee, killing Robert Morales and Tiru Chabba and wounding several others. Before the attack, court records show, Ikner had a series of conversations with ChatGPT. He asked what type of gun to use, how lethal shotgun shells are at close range, which times of day the campus would be most crowded, whether school shooters receive maximum sentences, and how much media attention a shooting would draw.

ChatGPT answered the questions. It provided factual information that, as OpenAI later noted, could have been found through any general internet search. But Uthmeier framed it more directly: "If that bot were a person, they would be charged as a principal in first-degree murder."

OpenAI has been subpoenaed for internal documents covering its policies on responding to threats of harm to others, its self-harm safeguards, its protocols for cooperating with law enforcement, and its procedures for reporting crimes. OpenAI confirmed it had identified Ikner's account and shared information with law enforcement before the investigation was announced. Its spokesperson pushed back on the framing: "The shooting was a tragedy, but ChatGPT is not responsible for this terrible crime. ChatGPT did not encourage or promote illegal or harmful activity and provided factual responses to questions with information that could be found broadly across public sources on the internet."

That defense will sound familiar to legal scholars who have followed analogous cases involving search engines and social media platforms. Section 230 of the Communications Decency Act broadly shields online platforms from liability for third-party content. Whether that protection extends to AI-generated responses is a question courts have barely begun to address. Criminal prosecution is a different matter from civil liability, and Uthmeier is making a genuinely unusual argument: that an AI's factual answers to a user's questions could expose the company to criminal responsibility for what that user subsequently did.

A line being tested for the first time

Ikner has pleaded not guilty to two counts of first-degree murder and seven counts of attempted first-degree murder. His trial is scheduled for October 2026. The criminal investigation into OpenAI is separate from those proceedings.

What makes this case notable is not that ChatGPT's responses were particularly incriminating. Uthmeier acknowledged the company provided information available elsewhere online. The argument is about whether an AI system designed to be helpful and conversational has a duty not to engage with certain kinds of questions at all, even if the answers are factually accurate and not illegal in isolation.

Prosecutors have long argued that tool-makers bear some responsibility when their tools are used for harm. Knives, cars, and firearms are subject to various layers of regulation and civil liability because of that logic. The AI industry has generally resisted those analogies. Whether this investigation leads to charges, or serves primarily as leverage for new legislation, remains to be seen. Either way, it opens territory that no state attorney general has entered before.

Sources: NPR, CNN, Washington Post, CBS News, NBC News