British Columbia has filed suit against OpenAI, alleging the company's safety staff flagged a teenager's violent ChatGPT prompts months before he killed nine people, including his mother and half-brother, and himself, in a February attack that also claimed five schoolchildren and one educator at Tumbler Ridge Secondary School Al Jazeera.

Attorney General Niki Sharma announced the province has retained CFM Lawyers in Vancouver and the California firm Stranch, Jennings & Garvey to press the claim in parallel with — but separate from — the litigation already brought by victims' families BC government. Sharma framed the action as a means of securing accountability from OpenAI and support for rebuilding the community, including construction of a new school facility, rather than as a substitute for the families' civil claims.

The underlying allegation is specific and damning. According to the province, OpenAI banned the account of 18-year-old Jesse Van Rootselaar in June 2025 — roughly eight months before the attack — after its systems flagged disturbing content involving the discussion and planning of violent scenarios. Twelve separate OpenAI employees reportedly urged the company to notify police Al Jazeera; the RPLE/Edelson filing makes the same claim RPLE & Edelson PC. No referral to law enforcement followed. OpenAI has said publicly that the account's activity did not meet its internal threshold of "an imminent and credible risk of serious physical harm to others."

On February 10, 2026, Van Rootselaar killed his mother and half-brother before attacking Tumbler Ridge Secondary School, a remote mountain-community campus in northeastern BC. He died of a self-inflicted gunshot wound. Five children aged 11 to 13 and one educator were killed at the school; 27 others were wounded. The province's death toll of nine, encompassing the two family members killed beforehand, aligns with the most recent Reuters accounting Reuters.

The families of seven victims — five killed, two injured, among them survivor Maya Gebala, who was struck by three gunshots — filed their own federal suit roughly three months before BC's announcement, in the U.S. District Court for the Northern District of California, naming both OpenAI and CEO Sam Altman Reuters. That complaint, brought with attorney Jay Edelson, rests on a negligence theory: that the company owed a duty of care once its own safety apparatus identified the risk, and breached it by declining escalation The Guardian. OpenAI has denied liability, pointing to Van Rootselaar's documented history of mental illness as an intervening cause Reuters.

Altman took the unusual step of publishing a personal apology in the local Tumbler RidgeLines newspaper in late April, writing that he deeply regretted the company's failure to alert law enforcement before the attack Tumbler RidgeLines. The apology has not altered OpenAI's litigation posture; the company continues to contest the negligence claims in court even as its chief executive has conceded the substance of the allegation in public.

The province's decision to sue independently, rather than fold its interests into the families' action, signals an intent to pursue remedies a private tort claim cannot reach — infrastructure funding for a new school, and potentially broader findings on OpenAI's safety governance that could feed into regulatory policy. BC's claim proceeds through both a Vancouver-based team and Northern District of California counsel, mirroring the cross-border structure the families' lawyers already established, since OpenAI is headquartered in San Francisco and any US-based liability findings will likely run through California courts regardless of where the harm occurred.

The case has already reshaped Ottawa's posture toward AI safety regulation. Canadian officials summoned senior OpenAI representatives to Ottawa in late February for urgent talks on the company's safety protocols, days after the shooting Reuters. Officials emerged from that meeting expressing disappointment that OpenAI presented no substantial new safety measures Reuters, and within days the federal government warned the company to strengthen its safeguards voluntarily or face government-mandated requirements Reuters. That threat materialized in June, when Ottawa introduced chatbot-regulation legislation explicitly tied to the Tumbler Ridge case, though the bill's prospects remain contested amid industry pushback Reuters.

What distinguishes this case from earlier disputes over AI-generated harm is the alleged gap between internal detection and external action. Plenty of litigation has turned on whether a platform's content moderation was adequate; fewer cases allege that a company's own safety team correctly identified a specific, named threat and was overruled by leadership on notification. If the province's pleadings hold up under discovery, the dispute over where AI companies' duty to warn begins and ends will move from academic debate to binding precedent, with consequences reaching well beyond one province or one company.