Investigators are examining exchanges between the shooter and OpenAI’s chatbot, as the company pledges to strengthen its safeguards.
Canadian investigators turned to ChatGPT as part of the probe into the Tumbler Ridge shooting in British Columbia after establishing that the perpetrator had interacted with OpenAI’s chatbot prior to carrying out the attack. This lead, confirmed following technical checks, places artificial intelligence at the center of the inquiry without establishing a direct causal link.
The tragedy, which occurred on February 10, has deeply shaken the country. The 18-year-old woman first killed her mother and half-brother before going to a secondary school, where she shot dead five students and a teaching assistant, then took her own life.
According to police, the perpetrator had a history of mental health issues. In a country where school shootings remain extremely rare, the shock has been profound.
Authorities revealed that the shooter’s account had been suspended by OpenAI in June 2025 after reports of concerning content, including scenarios involving armed violence.
At the time, the company had determined that the exchanges did not meet the threshold requiring notification of law enforcement.
Summoned to Ottawa, OpenAI representatives met with Canada’s Minister responsible for Artificial Intelligence, Evan Solomon. In an official letter, the company announced “immediate measures,” including strengthening security protocols, improving cooperation with federal and provincial authorities, and enhancing mechanisms to detect high-risk behavior.
OpenAI also stated that it has partnered with mental health and behavioral science experts to refine its criteria for assessing conversations that may constitute an “imminent and credible risk.” The company says it aims to better distinguish between fictional or emotional expression and statements that could signal a genuine threat.
The company’s response comes amid criticism. Several Canadian political leaders have expressed disappointment over the initial lack of concrete measures presented during the Ottawa meeting.
The Tumbler Ridge shooting, the second deadliest school attack in Canadian history after the 1989 École Polytechnique massacre in Montreal, has reignited debate over the responsibility of technology platforms in detecting early warning signs of violence and the limits of artificial intelligence tools.
