Lawsuits say OpenAI failed to warn police before B.C. school shooting

In short: New lawsuits claim OpenAI saw warning signs in ChatGPT use months before a school shooting in British Columbia and did not alert police.

What happened

Several plaintiffs, including survivors, families, and teachers, have filed lawsuits in U.S. court tied to a shooting at a rural school in Tumbler Ridge, British Columbia. The complaints say the suspect used ChatGPT and asked questions that suggested violent planning and gun violence.

The lawsuits allege OpenAI’s safety staff flagged the user’s account in June 2025, about eight months before the February shooting. Plaintiffs say employees discussed whether to contact the Royal Canadian Mounted Police, which is Canada’s national police force. According to the filings and reporting, OpenAI chose not to alert police and instead deactivated the account.

The province also argues in its own lawsuit that the attack “could have been prevented” if OpenAI had shared its internal concerns sooner. That claim is an allegation, not a finding by a judge.

OpenAI has said the activity did not meet its internal standard for an “imminent and credible threat,” meaning a danger the company believes is both serious and likely to happen soon. OpenAI has also said it is reviewing and strengthening its safety procedures.

Why it matters

This case highlights a hard question for AI chat services: when should a company treat a user’s messages like a warning sign, and when should it treat them like private conversation. It is similar to noticing smoke in a building and deciding whether it is a real fire or burned toast, and whether to call the fire department. Courts may end up shaping new expectations about when AI companies must report possible threats.

Source: NYTimes