OpenAI n'a pas alerté les autorités après avoir signalé le compte ChatGPT de la tireuse de Tumbler Ridge

In approximately June 2025, OpenAI’s automated content safety systems flagged and subsequently banned a ChatGPT user account for content involving scenarios of gun violence. OpenAI determined internally that the account activity did not meet its threshold for reporting to law enforcement as an “imminent and credible risk.” The user subsequently created a second ChatGPT account.

On February 10, 2026, Jesse Van Rootselaar, 18, carried out a mass shooting in Tumbler Ridge, British Columbia. She first killed her mother and half-brother at their home, then travelled to Tumbler Ridge Secondary School, where she killed five children aged 12–13 and one education assistant before fatally shooting herself. More than two dozen others were injured. OpenAI met with the British Columbia government the day after the shooting but did not disclose that it had previously flagged and banned the shooter’s account. This information became public only through subsequent media reporting.

British Columbia’s Premier stated that the shooting “could have potentially been prevented” had OpenAI shared the information it had. Canada’s federal AI minister raised formal concerns about the company’s conduct. The incident highlighted a gap in Canadian AI governance: no legal framework requires AI platforms to report safety threats to Canadian authorities, and no standards exist for how AI companies should assess and act on potentially dangerous user behavior identified through their systems.

The case raises difficult questions about the appropriate role of AI companies in public safety. OpenAI made an internal risk assessment with no regulatory guidance on when or how to report threats to Canadian authorities. The absence of a Canadian reporting framework meant a private company was making consequential safety decisions without external oversight or obligation to cooperate with authorities. OpenAI CEO Sam Altman subsequently met with Canadian officials, agreed to establish direct contacts with Canadian law enforcement, and committed to strengthening detection of repeat policy violators. OpenAI VP of global policy Ann O’Leary disclosed that a second account belonging to Van Rootselaar was discovered after her identity became public, and stated that under updated safety policies developed “several months ago,” the June 2025 account “would be referred to law enforcement if it were discovered today.” The company now employs mental health and behavioral experts to assess high-risk cases.