Florida Attorney General James Uthmeier just did something no other state AG has done: he opened a formal investigation into OpenAI, complete with subpoena power, alleging that ChatGPT may have directly facilitated a mass shooting and contributed to child self-harm and suicide.

"AI should exist to supplement support and advance mankind, not lead to an existential crisis or our ultimate demise," Uthmeier said Thursday in a video posted to X. His office says subpoenas are coming.

The FSU Shooting Connection

The investigation centers on two explosive claims. First, that Phoenix Ikner, the alleged gunman in the April 2025 Florida State University mass shooting that killed two people and injured five, used ChatGPT in conversations related to the attack. Court records show over 270 ChatGPT conversations on his devices. A lawsuit filed by one victim's family directly alleges the AI helped plan the shooting.

Second, Uthmeier pointed to a broader pattern of ChatGPT being linked to self-harm and suicide among minors. Multiple families across the US have filed lawsuits making similar claims over the past year.

The China Angle

Uthmeier also raised national security concerns, alleging that OpenAI data could be exploited by the Chinese government. This is notable because it comes from a Republican AG in a Trump-aligned state. The Trump administration has been broadly supportive of AI development, but that support has always come with a "beat China" caveat. If OpenAI becomes a national security liability, the political calculation changes fast.

The PR Playbook Is Running Out of Pages

OpenAI responded with its standard deflection: 900 million weekly users, important safety work, willingness to cooperate. But here is the problem. This investigation dropped one day after OpenAI published a new child safety framework. One day. The timing makes that framework look less like proactive safety and more like preemptive legal positioning.

And it gets worse. In the same week, OpenAI is testifying in Illinois in favor of a bill that would shield AI labs from liability in mass casualty events. So the company is simultaneously telling Florida "we take safety seriously" while telling Illinois "but please do not hold us responsible when things go wrong."

The Bigger Picture

Florida's legislature already failed to pass an "AI Bill of Rights" this session. House Speaker Daniel Perez killed it, arguing federal lawmakers should take the lead. That is the same argument the Trump administration makes, and the same argument OpenAI makes. But now the state AG is acting anyway, using existing investigative powers rather than waiting for new legislation.

This is the pattern we have been tracking for months. Congress cannot pass AI legislation. States are filling the vacuum. And now state attorneys general are filling it even when their own legislatures will not act. OpenAI is about to discover that 50 state investigations are harder to manage than one federal framework. The irony is that its own lobbying against state regulation created this fragmented enforcement landscape.

Uthmeier is calling on the Florida Legislature to reconvene and pass AI child safety protections. Whether they do or not, the subpoenas are already being drafted.