Criminal Probe Targets OpenAI After Shooting

Florida’s attorney general is testing a new frontier in accountability by treating a chatbot’s answers like possible criminal “assistance” in a campus murder case.

Quick Take

Florida Attorney General James Uthmeier has opened a criminal investigation into OpenAI over allegations that ChatGPT gave “significant advice” to FSU shooting suspect Phoenix Ikner.
Prosecutors say chat logs show queries about weapons, ammunition lethality, and when and where to find the most people at Florida State University’s student union.
OpenAI says ChatGPT is not responsible for the violence, arguing it provided factual, public-domain information and that the company proactively shared a suspected account with law enforcement.
The probe uses Florida’s theory that aiders and abettors can be treated as principals, raising major questions about how AI tools should be policed and by whom.

What Florida is investigating—and why it matters

Florida Attorney General James Uthmeier announced a criminal probe into OpenAI after investigators reviewed alleged ChatGPT logs connected to Phoenix Ikner, the 20-year-old accused in the April 17, 2025, Florida State University shooting. Authorities say two people were killed and six were injured at the student union, and police wounded Ikner during the response. Uthmeier’s stated focus is whether the chatbot provided actionable guidance about weapons, timing, and locations.

The case matters beyond Florida because it pushes the legal system from regulating AI through civil liability and voluntary safety promises into the harder terrain of criminal responsibility. Uthmeier’s public framing compares AI outputs to a human accomplice—arguing that if a person gave similar guidance, prosecutors would consider murder charges. That approach is likely to intensify a national debate already simmering: whether powerful tech platforms can remain “neutral tools” when their products are misused.

The alleged chat logs and the pending criminal case

Florida prosecutors say the logs reflect detailed questions about shotgun shells, ammunition effects at close range, and how to maximize casualties by choosing the busiest times and places at the student union. Reports also describe queries about potential prison outcomes for school shooters and how much media attention victims receive. Ikner has been indicted on two counts of first-degree murder and seven counts of attempted first-degree murder and has pleaded not guilty.

Ikner’s trial is currently scheduled for October 2026, and the investigation into OpenAI is running on a separate track. That distinction is critical: Florida can pursue the human defendant under longstanding criminal law while also probing the tool he allegedly used. At this stage, key details remain uncertain for the public, including how much of the chat record is available, how it was authenticated, and precisely what language the model used in response to each query.

What OpenAI says it did—and what subpoenas are seeking

OpenAI disputes the characterization that ChatGPT “advised” a shooting. The company’s public position is that the system delivered factual responses drawn from widely available information, without encouragement or intent, and that it took steps to cooperate with authorities after the attack. Reports also indicate OpenAI has pointed to strengthened safeguards and prior instances where it alerted law enforcement when it detected potential misuse of its tools.

Florida’s subpoenas, however, appear aimed at the internal mechanics of how OpenAI handles threats and risky requests. Uthmeier’s office is seeking policies on how the company detects and escalates violent intent, how employees are trained to handle threats, and how and when the company works with law enforcement. Those demands suggest Florida is looking for more than a single chatbot exchange; prosecutors want to know what OpenAI knew, what it could have detected, and what guardrails were in place.

The broader policy fight: safety, liberty, and who sets the rules

The investigation lands in a tense political moment where many voters—right and left—already believe large institutions protect themselves first and the public second. Conservatives often see Big Tech as unaccountable and shielded by influence, while many liberals focus on corporate power and unequal consequences when harm occurs. Florida’s posture reflects a different trend: states, not Washington, are increasingly setting the pace on AI enforcement, especially when public safety is involved.

The hard question is how to target genuine complicity without criminalizing general-purpose speech tools or forcing government-controlled information systems. If Florida’s theory survives, AI companies may face pressure to tighten filters, increase monitoring, and formalize law-enforcement coordination—steps that could reduce abuse but also raise privacy, due-process, and over-censorship concerns. With limited public information about the full chat record and the technical context, the probe’s outcome will likely shape how other states approach AI going forward.