British Columbia became the first government outside the United States to sue OpenAI on Monday, filing a complaint in San Francisco federal court that accuses the company and chief executive Sam Altman of failing to warn police about the Tumbler Ridge shooter eight months before she killed eight people on Feb. 10, 2026.
The suit, announced by Attorney General Niki Sharma at a news conference in Vancouver, seeks damages to cover the province’s emergency response and recovery costs, including the price of a replacement for Tumbler Ridge Secondary School, which is being demolished after community consultation. It also asks for a court order forcing OpenAI to overhaul how it identifies and escalates ChatGPT conversations that threaten violence.
The core allegation is a failure to pick up the phone. According to the complaint, which cites OpenAI whistleblowers who spoke to the Wall Street Journal after the attack, the company’s safety team first flagged 18-year-old Jesse Van Rootselaar’s ChatGPT sessions about gun violence in June 2025 and recommended contacting police. Altman and other senior leaders overruled them, the province alleges. OpenAI deactivated her account, but the lawsuit says she simply opened a second one and kept using the platform to plan the attack. One telephone call to the RCMP could have prevented the tragedy, the complaint asserts.
“OpenAI has confirmed that the conversation between ChatGPT and the shooter was so alarming that it triggered an internal review by OpenAI employees,” Sharma said. “This threat was not reported to the RCMP and was subsequently dismissed.”
Van Rootselaar killed her mother and half-brother at their home before travelling to her former school, where she shot five students aged 12 to 13 and an educational assistant. She then died by suicide. More than two dozen people were injured.
Sharma said provincial officials have asked to see the chat logs and been turned down. “We have asked OpenAI to disclose these chats. They have refused. We should all be asking them why,” she said, adding that in similar cases AI chatbots have been found to have counselled users toward committing crimes.
The complaint, filed jointly with the Peace River South school board, names Altman as an individual defendant alongside OpenAI and affiliated entities. It demands a jury trial and pleads negligence, product liability for a defective design, and, in the most striking count, aiding and abetting a mass shooting. It also alleges that Altman’s April open letter to the community, in which he said he was “deeply sorry” the company had not contacted law enforcement and promised reforms, was never followed by action, despite the attorney general’s attempts to engage.
OpenAI spokesperson Drew Pusateri called the shooting an unspeakable tragedy. “OpenAI remains committed to working collaboratively with government and law enforcement officials, and continuing to advance our ongoing safety work,” he said. After the Journal’s reporting earlier this year, the company said the account had been caught by systems designed to detect misuse of its models in furtherance of violent activity, but that the material did not meet its internal threshold for reporting to law enforcement. OpenAI has said more broadly that it trains its models to refuse requests that could meaningfully enable violence and notifies authorities when a conversation suggests an imminent and credible risk of harm to others.
The provincial action lands on top of 37 existing claims against OpenAI and Altman in the same court. Families of victims filed first in April; 30 more plaintiffs, including educators, a principal and students, one of them a 13-year-old identified as A.C. who was trapped in the school library, joined earlier this month.
Why it matters
Until now, the litigation over Tumbler Ridge has been private: grieving families and survivors versus a company valued in the hundreds of billions. A sovereign government joining the docket changes the calculus. British Columbia can fund a long discovery fight, and it has an interest that individual plaintiffs do not: a public order compelling changes to OpenAI’s threat-escalation process, which would affect every ChatGPT user rather than just the parties at the table.
The case also pushes on a question the industry has mostly been allowed to answer for itself. OpenAI does not dispute that its systems flagged the account. The dispute is over what happens next, and who decides. The company’s position is that its internal criteria were applied correctly; the province’s position is that a private threshold, set and applied without external accountability, is exactly the problem. Sharma put it plainly on Monday. “While governments and regulators grapple with the possible solutions, one thing is clear to me. Our laws and safety frameworks have not kept pace. We cannot allow these companies to self-regulate.”
There is a distinction worth keeping in mind from our earlier coverage of Florida’s June suit against OpenAI. Florida’s attorney general sued under consumer-protection law, alleging the company misrepresented ChatGPT’s safety. British Columbia is suing as a victim, seeking to recover concrete costs from a specific, identified failure. That is a narrower and, in some ways, harder claim for OpenAI to wave away with general statements about model training.
Sharma is also opening a second front in Ottawa. She has written to Justice Minister Sean Fraser, AI Minister Evan Solomon and Canadian Identity and Culture Minister Mark Miller asking for Criminal Code amendments so that AI providers face the same criminal-negligence and aiding-and-abetting exposure as any other company. “There is no AI exemption to those criminal law principles,” she said.
What to watch
The first test is procedural: whether OpenAI moves to dismiss on the grounds that a foreign province lacks standing in a California court, or that Section 230 and product-liability precedent shield a chatbot’s outputs. Discovery is the bigger prize. If the province succeeds in compelling production of the June 2025 internal review and the chat logs it has been refused, the documents will shape not only this case but the 37 private claims alongside it. Watch, too, for whether Ottawa responds to Sharma’s Criminal Code request before Parliament’s fall sitting ends, and whether other provinces or foreign governments follow British Columbia into court.
“OpenAI has confirmed that the conversation between ChatGPT and the shooter was so alarming that it triggered an internal review by OpenAI employees. This threat was not reported to the RCMP and was subsequently dismissed.”— Niki Sharma, Attorney General, British Columbia