British Columbia’s lawsuit connects OpenAI to the Tumbler Ridge school shooting through an alleged missed intervention: the province says OpenAI had warning signs in the shooter’s ChatGPT activity, suspended an account, but did not alert the RCMP. The case, filed in federal court in San Francisco on September 21, names OpenAI and CEO Sam Altman. It remains unresolved, so the province’s allegations are not judicial findings.
17
19
What happened at Tumbler Ridge
On February 10, 18-year-old Jesse Van Rootselaar killed eight people—six children and two adults—in Tumbler Ridge, British Columbia, before dying from a self-inflicted gunshot wound, according to reporting on the province’s lawsuit.
22
The provincial government argues that the attack is relevant to OpenAI’s product design and threat-reporting practices because the shooter allegedly used ChatGPT before the violence. Its claim is not merely that ChatGPT existed in the background, but that OpenAI had information serious enough to warrant action beyond closing an account.
17
22
British Columbia’s core allegation
B.C. alleges that OpenAI negligently designed and distributed an unsafe product and failed to notify law enforcement about threats made through ChatGPT before the shooting. The province says a warning could have reached local police and prevented the attack. It is seeking compensation for past and future public recovery costs, as well as a court order requiring changes to how OpenAI identifies and handles ChatGPT conversations that may lead to violence.
17
25
That proposed chain of events—what police would have known, what they could have done, and whether intervention would have prevented the attack—is central to the litigation. It will need to be tested through evidence and legal argument rather than assumed from the existence of a flagged account.
The disputed June 2025 account review
OpenAI has acknowledged that it identified an account associated with Van Rootselaar in June 2025 through its abuse-detection and enforcement systems, which include efforts to identify use of models in furtherance of violence. The company banned the account.
33
OpenAI says it did not report the account to authorities because the information available at the time did not show a credible and imminent plan for serious physical harm—the company’s threshold for law-enforcement referral.
33
43
B.C.’s suit disputes the adequacy of that response. The province’s position is that an internal review and account ban were warning signs that should have led OpenAI to contact the RCMP. OpenAI has since said that, under security protocols adopted after June 2025, it would have notified police about the account.
39
44
Why the chat history matters
The public record does not contain the full ChatGPT exchanges. Attorney General Niki Sharma said OpenAI declined to provide the shooter’s chat history to the province. That limits the public’s ability to independently assess the conversations’ specificity, the company’s internal decision-making, and the realistic prospects for intervention had police been notified.
37
This missing detail also explains why the dispute should be described carefully: the legal claims concern alleged failures to design, detect, escalate, and prevent harm, rather than proven conclusions about exactly what the chatbot said or caused.
Related lawsuits and the account-bypass issue
Families of victims, survivors, students, and educators have filed related actions in California. OpenAI has sought dismissal of more than three dozen cases, arguing in part that claims over deaths and injuries in British Columbia should be litigated there rather than in California.
26
28
Separate claims also raise the question of whether a ban alone is an effective safety measure. OpenAI has said the shooter created a second account after the June 2025 suspension and that it was unaware of that account until after the shooting. The company has said it is making it harder for banned users to return to the service.
36
45
What OpenAI and Altman have said
OpenAI has expressed sorrow over the attack, said it has a zero-tolerance policy for use of its tools to facilitate violence, and said it has strengthened safeguards. Reporting on the later litigation also described an apology from Altman.
38
39
Its central legal and factual position is narrower: banning the account in June 2025 was appropriate based on the information it had, and the activity did not meet its then-current standard for reporting an imminent and credible threat to law enforcement.
33
43
What the province is seeking
B.C. is seeking damages tied to emergency response and community recovery, including costs associated with replacing the school, alongside changes to OpenAI’s practices for potentially violent use of ChatGPT.
17
23
The lawsuit therefore poses a broader question for AI companies: when a platform flags violent activity but lacks certainty that an attack is imminent, what duty—if any—does it have to report that activity to law enforcement? The court has not yet answered that question in this case.