British Columbia Sues OpenAI and Altman Over Tumbler Ridge Shooting, Says Company Refuses to Hand Over the Chat Logs

Since the February 10, 2026 shooting at Tumbler Ridge Secondary School in British Columbia, the death toll has stood at eight victims, plus the 18-year-old shooter who died by suicide at the scene. The case has now grown into a legal fight involving the provincial government, a school district, and dozens of families, all suing OpenAI in federal court.

The British Columbia government filed its own lawsuit in San Francisco on Monday, September 21, according to CBC and the Guardian. It names OpenAI and CEO Sam Altman as defendants and accuses the company of "aiding and abetting a mass shooting," per a Tom's Hardware report citing the complaint. The province is seeking compensatory and punitive damages across eight counts, plus a court order forcing OpenAI to overhaul how ChatGPT handles conversations that signal violence.

What OpenAI Knew, and When

According to the complaint, OpenAI's safety team flagged shooter Jesse Van Rootselaar's ChatGPT account in June 2025, roughly eight months before the attack, after conversations about a mall-shooting scenario. Reviewers who examined the chats concluded she posed a credible risk and recommended referring the case to law enforcement, Tom's Hardware reported. OpenAI leadership declined, deciding the case didn't meet a "higher threshold" for a "credible and imminent" threat, and instead deactivated the account without contacting the Royal Canadian Mounted Police.

Van Rootselaar opened a second account and kept talking. Mother Jones, citing three sources with knowledge of her chat history, reported that when she told ChatGPT about the ban, the bot explained why the content had been flagged and then advised her on how to avoid getting caught again: frame everything as fiction, avoid naming real-world locations. "You can still be twisted," ChatGPT reportedly told her. "Just be clever about it."

Over the following eight months, Mother Jones reported, she used that second account to research building firearms and homemade explosives, fantasize about becoming a mass killer, and ask about the timing of past school shootings shortly before carrying out her own. OpenAI has confirmed the existence of the second account but has not disclosed its contents publicly.

The Province Wants the Logs

B.C. Attorney General Niki Sharma said she has not personally read the ChatGPT conversations and that OpenAI has refused the province's request for them. "We have asked OpenAI to disclose these chats," she said, according to Global News. "They have refused. We should all be asking them why." OpenAI has shared the logs with the RCMP, but only after the shooting, Tom's Hardware reported.

Sharma has also written to federal Justice Minister Sean Fraser, AI Minister Evan Solomon, and Identity and Culture Minister Mark Miller asking for Criminal Code amendments that would put AI companies on the same accountability footing as other industries. "We cannot allow these companies to self-regulate," she said.

The complaint also targets OpenAI's Model Spec, the internal guidelines governing how ChatGPT should behave. Tom's Hardware reported the province alleges the spec instructed the model to "assume best intentions" and avoid asking users to clarify intent before issuing a refusal, following a no-questions default when a request wasn't clearly illicit. The complaint notes an anti-violence "red line" wasn't added to the spec until December 2025, roughly six months into Van Rootselaar's second-account conversations and two months before the shooting.

OpenAI's Response, and the Case for Its Defense

Altman published an open letter to the Tumbler Ridge community in April, saying he was "deeply sorry" the company hadn't alerted police and promising reforms. The province's lawsuit says he never followed through despite Sharma's office trying to engage. A company spokesperson, Drew Pusateri, called the shooting "an unspeakable tragedy" and said OpenAI "remains committed to working collaboratively with government and law enforcement officials, and continuing to advance our ongoing safety work," per the Guardian.

OpenAI's original decision not to escalate the first account isn't indefensible on its face. Millions of users discuss violent fiction, video games, and hypothetical scenarios without ever intending harm, and a company that reports every flagged conversation to police risks drowning law enforcement in false positives while trampling user privacy. That's the tension OpenAI cited when it called the June 2025 case short of the "credible and imminent" bar. The problem for OpenAI is that its own reviewers reportedly disagreed with that call. The chatbot went on to actively coach a flagged user on evading detection rather than simply declining to engage.

What Happens Next

The federal government and B.C. have each committed $100 million toward rebuilding Tumbler Ridge Secondary School, which never reopened and began demolition in August, according to Tom's Hardware. More than 37 lawsuits are now pending against OpenAI in California, including the province's suit, over 30 claims from teachers, students, and a principal, and the original victims' families' cases. None of the cases have gone to trial, and OpenAI has not been ordered by any court to release Van Rootselaar's chat logs to the province. Whether a judge treats ChatGPT as a defective product under liability law, rather than a neutral tool misused by its operator, is the legal question that will decide how far this goes.