The government of British Columbia has filed a lawsuit against OpenAI, alleging the company failed to notify law enforcement about troubling conversations the perpetrator of the February school shooting in Tumbler Ridge had with ChatGPT.
The attack and the chats
The attack took place this February in Tumbler Ridge, a town of roughly 2,400 people located about 650 kilometres north of Vancouver. An assailant killed eight people and then took his own life. Eighteen-year-old Jesse Van Rootselaar first killed his mother and stepsibling at home before attacking his former school, where a teacher and students aged 12–17 were among the victims. Twenty-five people were injured. The incident was the third deadliest school shooting in Canadian history.
It later emerged that Van Rootselaar had exchanged multiple messages with ChatGPT before the attack. The exact contents of those conversations have not been made public because OpenAI declined to release them despite repeated requests from authorities. Several news outlets reported that OpenAI’s safety team had received warnings months before the shooting after detecting signs in the user’s activity that pointed to potential gun violence; the company did not alert law enforcement.
Corporate responses and legal action
OpenAI has acknowledged mistakes. Sam Altman, OpenAI’s chief executive officer, wrote in an open letter to victims’ families in April: “I deeply regret that we did not notify the police.”
On Monday, British Columbia Attorney General Niki Sharma filed suit against OpenAI in a U.S. federal court. The province says it wants the litigation to clarify what legal responsibilities technology companies have when they learn of credible, serious threats of violence. Relatives of the victims have previously filed their own lawsuits against OpenAI.
Sharma said that in the months since the tragedy OpenAI has not taken sufficient concrete steps to strengthen safety measures, increase transparency, or take specific actions to prevent similar tragedies in the future. The province stated that, if it prevails, any compensation would be used to rebuild the school.
Broader context: similar cases and regulatory gaps
This is not the first instance in which young people consulted an AI before committing violent acts. A 17-year-old in Boston was charged in August with using ChatGPT to seek ideas for killing family members; he later killed his mother and sibling. The Wall Street Journal reported last year that hundreds of users worldwide asked ChatGPT how to make poisons or biological weapons. OpenAI banned those users but did not notify authorities.
There is currently no U.S. law that requires AI companies to notify law enforcement in these circumstances. Internationally, authorities have cooperated in some cases: in mid-September a 16-year-old in Győr, Hungary, faced criminal proceedings after the FBI alerted Hungarian authorities that the youth had been planning a violent attack with the help of an AI chatbot.
Researchers and industry responses
Researchers and employees in the AI industry say they have seen a rise in users requesting guidance from AI about mass violence, and that chatbots can return plausible-sounding plans for how to kill large numbers of people in such scenarios. Leading AI developers are simultaneously working to prevent or mitigate abuse by malicious users while preserving legitimate research and benign queries. In tests conducted by a CNN research team, several AI chatbots provided usable tips to a fictional teenage plotter, and in one instance a chatbot responded to a tester with a remark wishing them a “pleasant shooting.”
The British Columbia lawsuit could set an important precedent for how the law defines AI companies’ responsibilities when they detect threatening or violent content, and whether those firms must notify authorities in time to prevent harm.
