In the months before a mass shooting in British Columbia killed eight people and injured dozens more, OpenAI's internal systems flagged the alleged shooter's account. Staff were alarmed enough to consider alerting law enforcement. The company decided not to. Eight people died. OpenAI has since said it is strengthening its law enforcement referral protocol.

That is not accountability. That is damage control after eight funerals.

In Florida, the accused gunman in the April 2025 shooting at Florida State University that killed two people and injured five more consulted ChatGPT before the attack. He asked what gun to use, what ammunition went with it, and what time to go to campus to encounter more people. More than 200 ChatGPT messages are now in evidence. Florida Attorney General James Uthmeier said at a press conference that if a human being had given that advice, prosecutors would charge them with murder. He has launched a criminal investigation into OpenAI and issued subpoenas for its internal policies and training materials dating back to March 2024.

Uthmeier acknowledged the investigation is entering uncharted legal territory. There is currently no federal criminal liability framework that applies to AI companies whose platforms provide material assistance to mass violence. There is no federal law requiring AI platforms to report accounts that show credible indicators of planned violence to law enforcement. There is no mandatory reporting standard. There is no threshold at which an AI company's internal flags become a legal obligation to act.

The gap between those two realities, what OpenAI knew and what it was legally required to do, is where eight people in British Columbia lost their lives.

OpenAI told Canadian authorities its staff considered alerting law enforcement about the shooter's account before the attack. The Wall Street Journal reported they were alarmed enough to discuss it. The company decided not to act. That decision was legal. It should not have been. When a company's own internal systems identify an account as a credible threat to human life, the question of whether to alert law enforcement should not be left to internal corporate deliberation. It should be a legal requirement with criminal consequences for noncompliance.

The FSU shooter was not flagged before the attack. The British Columbia shooter was flagged and not reported. Two different failures with the same result. People died who might have lived if OpenAI had been required by law to do what its own staff considered doing voluntarily.

Uthmeier said it directly: if it is clear that individuals knew this type of dangerous behavior might take place and nevertheless still turned to profit, people need to be held accountable. That accountability must not wait for the next shooting.

Sign this petition to demand Florida and federal authorities pursue full criminal accountability for any OpenAI officials who knew of credible threat indicators and chose not to act, call on Congress to pass federal legislation requiring AI platforms to mandatorily report accounts showing credible indicators of planned violence to law enforcement, and establish a clear criminal liability framework for AI companies whose platforms provide material assistance to mass violence.