British Columbia has filed a federal lawsuit in San Francisco against OpenAI and CEO Sam Altman, alleging the company's internal safety team identified Jesse Van Rootselaar as a threat months before she killed nine people — mostly children — at a school in Tumbler Ridge on 10 February 2025, but that company leadership overruled a recommendation to contact police. The province is seeking damages and a court order forcing changes to how OpenAI handles violent content flagged in ChatGPT conversations. The factual spine of the case is damning in its specificity. According to the lawsuit, which cites OpenAI whistleblowers who spoke to the Wall Street Journal, safety team members flagged Van Rootselaar's conversations about gun violence in June 2025 and recommended contacting law enforcement. Altman and other senior leaders vetoed that recommendation. The shooter's account was deactivated, but she simply created a new one and continued planning. OpenAI later said the conversations did not meet its internal threshold for reporting to police. This is the structural question the lawsuit forces into the open: who decides where the reporting threshold sits, and who bears the cost when that threshold is set too high? OpenAI designed its own criteria for when to alert law enforcement. Its own safety team judged those criteria were met. Its own leadership overruled that judgment. Nine people, primarily children aged 12 and 13, are dead. The company's internal governance failed on its own terms. The province's legal strategy is notable for what it demands beyond money. BC wants a court-ordered restructuring of OpenAI's content monitoring and reporting protocols — essentially asking a federal judge to impose external governance on a private company's safety decisions. This goes well beyond the individual family lawsuits already filed by more than 30 plaintiffs. It is a sovereign government asserting that the current self-regulatory regime is inadequate. Altman published a letter in April to the Tumbler Ridge community saying he was "deeply sorry" and promising reforms. According to the lawsuit, those reforms never materialized despite outreach from BC's attorney general. The gap between public contrition and operational change is the extraction mechanism here — reputational repair without structural cost. The case lands in an accelerating legal environment. Florida became the first US state to sue OpenAI over safety risks in June. Separate lawsuits link ChatGPT to a Florida State University shooting. Canadian officials called OpenAI executives in for questioning, and Canada's AI minister spoke directly with Altman. The company's legal exposure is compounding across jurisdictions, but no court has yet established a duty-to-warn standard for AI platforms. OpenAI's stated position — that it trains models to refuse requests that could "meaningfully enable violence" and notifies law enforcement when conversations suggest "an imminent and credible risk of harm" — now confronts evidence that its own safety employees believed those conditions were met and were overridden. The distance between the company's public safety claims and its internal decision-making is the core of every lawsuit it now faces.
