Sovereign AI Product Liability: British Columbia Sues OpenAI in U.S. Federal Court Over Tumbler Ridge

Sovereign AI Product Liability: British Columbia Sues OpenAI in U.S. Federal Court Over Tumbler Ridge

The sovereign-litigation track of the Tumbler Ridge mass-shooting legal wave has moved from announcement to filing. On September 21, 2026, the Province of British Columbia filed suit against OpenAI and CEO Sam Altman in the U.S. District Court for the Northern District of California (San Francisco), alleging OpenAI failed to notify authorities about concerning user interactions on ChatGPT ahead of the February 10, 2026 Tumbler Ridge mass shooting (eight killed by 18-year-old Jesse Van Rootselaar). Per Al Jazeera's account of the complaint, the lawsuit accuses OpenAI of "aiding and abetting a mass shooting" and product defect liability, seeks compensation for emergency response and community recovery costs — including rebuilding the school, which is to be demolished — and asks the court to force OpenAI "to revamp its policies for identifying and addressing user conversations that threaten violence."

The complaint's liability theory is framed as product design, not mere speech:

• "One telephone call to the RCMP could have prevented the tragedy"
• The platform is "not a passive conduit for the speech of others but a content-generating product that manufactured scenarios involving gun violence"
• "Feasible, safer design alternatives existed and were not adopted or effectively implemented"

The factual predicate: OpenAI flagged the shooter's account roughly eight months before the attack and deactivated it for "misuses of our models in furtherance of violent activities," but did not refer the matter to police because the content "did not meet the higher threshold required for referral." The Wall Street Journal reported that about a dozen OpenAI employees discussed whether to alert authorities; some urged referral, but leadership decided against it. British Columbia also alleges OpenAI failed to implement a user-level ban (the shooter opened a second account) and that ChatGPT's safety features were weakened to prioritize engagement. OpenAI says it shared chat logs with the RCMP; BC says it has not received them.

The forum irony is now the central procedural fact: British Columbia chose a U.S. court even as OpenAI moves to dismiss the 37 victims' suits on forum non conveniens grounds — arguing that litigation belongs in Canada. As Law.com put it: "British Columbia's lawsuit over a deadly mass shooting at Tumbler Ridge Secondary School comes as OpenAI has moved to dismiss dozens of lawsuits in California filed by the families of the victims on grounds that the litigation belongs in Canada.¹" Simon Fraser University's Jean-Christophe Belisle-Pipon observed BC is likely pursuing the U.S. forum because it "has a greater chance of recovering higher damages there."

Why this matters for enterprise risk: Drexel law professor Anat Lior frames the doctrinal stakes — if courts recognize a duty to warn, "it could impose affirmative obligations on AI companies, based on their exposure to user information, to alert authorities or take other protective measures on behalf of potential victims." A sovereign plaintiff with police-power resources suing a developer in a U.S. court, alongside this cycle's 42-state criminal-subpoena escalation (see prior revision), marks the shift of AI product liability from private tort claims to state-enforced litigation. Deployers of third-party AI should note the complaint targets both developer design choices (safety-feature tradeoffs) and incident-response duties (threat referral, user-level bans) — the same operational questions enterprises must answer in their own AI governance programs.