British Columbia sues OpenAI over Tumbler Ridge shooting

The suit also names CEO Sam Altman and says OpenAI failed to warn police after ChatGPT chats tied to the shooting.

SAN FRANCISCO / VICTORIA — British Columbia and School District No. 59 (Peace River South) sued OpenAI entities and CEO Sam Altman in the U.S. District Court for the Northern District of California, alleging the company failed to warn law enforcement after ChatGPT conversations tied to the Feb. 10, 2026 Tumbler Ridge mass shooting were flagged. Attorney General Niki Sharma said the province seeks accountability and stronger AI protections.

A school-district and Crown plaintiff ask a California federal court to treat a chatbot platform’s safety pipeline — flag, escalate, warn police or not — as a duty that can ground damages and court-ordered product changes.

The rule in English

Filing. Complaint (published PDF; case no. 3:26-cv-10743, San Francisco Division) lists claims including negligence (failure to warn law enforcement / failure to warn), negligent entrustment, aiding and abetting, negligent undertaking, negligence (design defect), and strict product liability (design defect / failure to warn). Jury demand.

Relief sought (Reuters / complaint coverage). Damages for provincial recovery costs (including replacement-school and mental-health burdens after the school is demolished); injunctive relief directing changes to how ChatGPT handles conversations that could lead to violence; urgent ask for disclosure of the shooter’s chat logs (per secondary complaint coverage).

Core allegation (pleadings). Safety systems flagged the shooter’s account for gun-violence-related misuse months before the attack; safety staff recommended contacting police; leadership overruled per whistleblower accounts cited in the complaint (WSJ-sourced in secondary coverage); account deactivated but a new account continued; OpenAI later said issues did not meet internal criteria for law-enforcement reporting. Altman publicly apologized for not contacting law enforcement and promised reforms; the province says those reforms were not followed through despite AG outreach (Reuters).

OpenAI response (Reuters). Spokesperson Drew Pusateri called the shooting an unspeakable tragedy; company says it remains committed to working with government and law enforcement and advancing safety work. Broader company position (prior cases): models trained to refuse requests that meaningfully enable violence; notifies law enforcement when conversations suggest imminent, credible risk of harm to others — disputed here as applied.

Related actions. Families already sued in California federal court. Florida earlier sued OpenAI over alleged ChatGPT safety failures involving children and violence (separate matter).

Who wrote it / who it binds

Plaintiffs: Crown (HMTK in right of B.C.) + SD59 Board of Education; AG Sharma leading for the province. Defendants: Altman personally + named OpenAI corporate entities. Binds no one until judgment or settlement. California federal pleading postures, not a Canadian criminal finding.

Who it exempts

Other chatbot vendors not sued in this complaint. Individual employees below the named defendants (unless later joined).

What a founder, parent, or worker should do differently

Platform founders: treat “flag but no LE notify” as the liability theory governments are now testing — document criteria, override logs, and law-enforcement escalation clocks.

Parents and schools: this is a civil accountability fight over whether a platform that sees planning signals must call police — not a how-to guide and not a substitute for local threat-reporting channels.

Policymakers: watch whether failure-to-warn + design-defect claims survive motions to dismiss — that shapes future mandatory-reporting bills for AI firms.