British Columbia sues OpenAI over warnings before mass shooting
The Canadian province of British Columbia announced a lawsuit against OpenAI , alleging the company internally flagged violent ChatGPT conversations linked to the person who carried out the Tumbler Ridge mass shooting but did not notify law enforcement. The February 10 attack killed eight people, including five children, and wounded 27.
According to the province, OpenAI banned the account in June 2025 for “disturbing content” involving violent scenarios, and 12 employees urged the company to alert police. Leadership declined, the suit says, concluding the activity did not represent “an imminent and credible risk of serious physical harm to others.” Attorney General Niki Sharma said the province wants to “hold OpenAI and its decision-makers accountable.”
The case tests an unsettled question: what duty an AI provider owes to third parties when its own monitoring surfaces a possible threat. OpenAI’s usage policies bar violent content and reserve the right to escalate imminent-harm cases to authorities, but the threshold for doing so is discretionary. A ruling that the company should have acted on its internal flags would push AI firms toward more aggressive reporting of flagged users — with attendant privacy and false-positive tradeoffs.