On September 21, the Canadian province of British Columbia sued OpenAI and its CEO Sam Altman in federal court in San Francisco, alleging the company failed to alert police to signs that the Tumbler Ridge school shooting was being planned through ChatGPT before the attack in February. It is the first time a provincial government has taken AI safety accountability to court.
What the lawsuit alleges
The core allegation: OpenAI's internal safety team had flagged the shooter's ChatGPT account as early as June 2025, found conversations involving gun violence in it, and recommended contacting law enforcement — but the company did not follow through. Attorney General Niki Sharma said in a statement: "The people of Tumbler Ridge deserve answers and accountability."
British Columbia is seeking damages for the costs of recovery, including mental health services and rebuilding the attacked school, as well as a court order directing OpenAI to change how it handles ChatGPT conversations that may indicate serious violence. The suit also says Altman apologized to the local community, admitting the police had not been alerted, but that the promised reforms never materialized. The province had said back in July it was exploring legal action; the suit was formally filed on September 21.
OpenAI's response
An OpenAI spokesperson called the Tumbler Ridge shooting an "unspeakable tragedy" and said the company "remains committed to working collaboratively with government and law enforcement officials, and continuing to advance our ongoing safety work." The company's position so far: its internal review at the time found no "credible and imminent planning" that met the threshold for referral to law enforcement — though under updated procedures, such cases would now be reported.
Notably, this is not the first lawsuit. More than 30 family members of victims have already sued OpenAI in California federal court, accusing the company of missing a chance to prevent the shooting. Individual accountability was already underway before the government stepped in.
Why this lawsuit carries different weight
Individual lawsuits pursue accountability case by case; a provincial government stepping in turns the question into a systemic one: when an AI company's internal safety processes are already flashing red, is there a legal duty to warn law enforcement? And who sets the threshold — the company itself, or the regulator?
The case also touches a thornier trade-off: the privacy of user conversations versus the public-safety need for warning. OpenAI processes enormous volumes of dialogue every day; flagging, escalation, and referral by the safety team each need clear rules. Rules too loose, and the cost of a missed signal can be lives; rules too strict, and false alarms erode user trust and create huge operational burdens. For the first time, a court will weigh this balance in an AI context.
What this means
Whatever the outcome, the suit draws a new accountability line for the whole industry: the records of internal safety reviews are becoming courtroom evidence. AI companies can no longer treat "we looked at it internally" as the finish line — what was done afterward, and what was not done and why, will be examined item by item.
For practitioners, the direct lesson: internal procedures for high-risk content involving violence or self-harm need auditable, explainable referral criteria — and they belong in publicly disclosed safety policies, not just internal manuals. For regulators, this lawsuit could accelerate legislative debate in many countries over a "duty to warn" for AI companies. Safety spending is turning, for the first time, from a cost item into a legal risk item.
