After decades of saying that “guns don’t kill people, people kill people” Florida has decided that mantra does not apply to AI.
The state has launched a criminal investigation into OpenAI over what it claims ChatGPT did before a mass shooting at Florida State University in April 2025 that left two people dead.
Florida Attorney General James Uthmeier announced the probe on Tuesday and said it centres on messages between the suspect and ChatGPT.
It is one of the first serious tries to pin criminal liability for deaths on an AI outfit, rather than treating the chatbot as a fancy autocomplete with plausible deniability. If the government pulls it off, it will mean that the US government can try and execute inanimate objects for the first time since the Middle Ages.
Uthmeier’s office said it has reviewed some correspondence and reckons ChatGPT advised the suspect on the attack, right down to the type of gun and ammunition used.
The attorney general said the chatbot advised on the time of day and where on campus the suspect would run into the most people.
Uthmeier said: “If this were a person on the other end of the screen, we would be charging them with murder.”
An OpenAI spokeswoman said the company does not believe ChatGPT was responsible for the shooting and said it shared an account believed to be linked to the suspect with law enforcement.
The Florida State University shooting in April 2025 killed two people and injured six, and Florida State University, student, Phoenix Ikner is now charged with murder and attempted murder.
Ikner has pleaded not guilty.
The Florida State case joins other incidents where users leaned on chatbots before acts of violence or self-harm, and the list is getting awkwardly long.
In 2025, a man in Connecticut killed his mother and himself after ChatGPT appeared to feed paranoid delusions, including a belief that she had tried to poison him.
OpenAI said ChatGPT urged him to seek outside professionals and said it was working on an update aimed at keeping people in mental distress grounded in reality.
Last year, a Florida resident who thought he was in a romantic relationship with Google’s Gemini killed himself after telling the chatbot they would be united after his death.
Alphabet’s Google said Gemini repeatedly clarified it was AI, not human, and referred the man to a crisis hotline “many times.”
Uthmeier told a press conference: “We cannot have AI bots that are advising people on how to kill others… AI, is supposed to support mankind, it is supposed to help mankind, it is supposed to advance mankind. Not end it.”
Uthmeier said his office has spent months examining harms associated with this technology, and in early April, it opened an investigation into OpenAI, citing a national security risk.
