For lawyers advising AI companies, the case raises a wider UKJT AI Liability Statement — Lawyer Monthly question about what follows once a platform’s internal safety systems identify potentially dangerous behaviour.
The lawsuit, filed in federal court in San Francisco, seeks damages for costs the province says it has incurred and will continue to bear following the shooting. British Columbia is also asking the court to order changes to the way OpenAI handles ChatGPT conversations that may indicate a risk of violence.
Eight victims were killed in the February 10 attack, including five students, an educational assistant and two members of the attacker’s family. The 18-year-old attacker subsequently died by suicide.
At the centre of the province’s case is what OpenAI allegedly knew before the shooting and how the company responded. According to the complaint, the attacker’s interactions with ChatGPT were first flagged in June 2025 after conversations concerning gun violence.
British Columbia alleges that members of OpenAI’s safety team recommended contacting police but that company leadership decided the material did not meet the threshold for referral to law enforcement. The account was deactivated, according to the lawsuit, but another account was later created and ChatGPT continued to be used.
OpenAI has said its systems had identified activity associated with the account as potential misuse connected with violent conduct, but that the material did not meet its internal criteria for reporting to authorities.
British Columbia Attorney General Niki Sharma said the province is seeking accountability over OpenAI’s alleged failure to notify law enforcement about threats made through its platform. The government says it has faced costs including replacement of the school, mental-health support and other measures required following the attack.
OpenAI spokesperson Drew Pusateri described the shooting as an unspeakable tragedy and said the company remains committed to working with governments and law enforcement while continuing to improve its safety systems.
The provincial lawsuit is separate from proceedings brought by more than 30 relatives of victims and others affected by the attack. Those cases, also filed in California federal court, make their own allegations against OpenAI, Microsoft Face Copyright Challenge | Lawyer Monthly arising from the same events.
For lawyers advising AI companies, the case raises a difficult question about what follows once a platform’s internal safety systems identify potentially dangerous behaviour. The issue is not simply whether an AI model generated harmful information, but whether knowledge obtained through monitoring and escalation procedures can create legal exposure if a company decides not to act.
That could make internal policies particularly significant. Records showing what triggered a safety review, who considered the material, what reporting threshold applied and why escalation was rejected may become important evidence when courts assess Gilead Duty-to-Innovate Ruling — Lawyer Monthly.
OpenAI says it trains its models to reject requests that could meaningfully facilitate violence and notifies law enforcement where conversations suggest an imminent and credible threat to others. British Columbia’s case will place greater scrutiny on how that threshold operates in practice and whether existing procedures were adequate in the circumstances alleged.
The litigation therefore goes beyond one disputed moderation decision. As AI platforms gain greater visibility into users’ conversations, courts may increasingly be asked to consider when knowledge of a potential threat can give rise to a legal duty and how responsibility should be allocated when an internal warning is identified but no external report is made.