Canada's Minister of Artificial Intelligence Evan Solomon summoned senior OpenAI safety officials to Ottawa last week to scrutinize the company's protocols following revelations about the ChatGPT account of Jesse Van Rootselaar, the suspect in the February 10 Tumbler Ridge school shooting.

Van Rootselaar, an 18-year-old resident of the remote British Columbia town, killed eight people before dying by suicide. The victims included her mother, 11-year-old half-brother, five young students, and a school educator. Two other students were injured in the attack, one of the deadliest school shootings in Canadian history.

OpenAI banned Van Rootselaar's first ChatGPT account in June 2025, more than seven months before the rampage, after identifying "misuses of our models in furtherance of violent activities," including troubling posts about gun violence scenarios. Internal employees flagged the activity as a potential indicator of real-world violence and urged leaders to notify Canadian police, but the company determined it did not meet the threshold for a credible or imminent plan of harm.

The suspect evaded the ban by creating a second ChatGPT account, which OpenAI only detected after police publicly named her. The company then shared details with law enforcement and reached out proactively to support the investigation.

Solomon described the reports as "deeply disturbing" and demanded explanations on OpenAI's safety protocols and escalation thresholds to the police. Federal officials, including Justice Minister Sean Fraser and Public Safety Minister Gary Anandasangaree, expressed disappointment after the meeting.

In response, OpenAI pledged several changes, including strengthened detection systems to prevent ban evasion, prioritization of high-risk offenders, enhanced referral protocols for imminent threats, and a direct point of contact with Canadian law enforcement. Company Vice President for Global Policy Ann O'Leary stated that under the updated rules, the June 2025 account would now be referred to the authorities.

British Columbia Premier David Eby questioned whether the tragedy could have been prevented had OpenAI warned police, noting prior mental health contacts and firearm removals from the home that were later returned. Van Rootselaar had posted online about issues including ADHD, depression, OCD, and autism spectrum traits.

Solomon plans to meet OpenAI CEO Sam Altman next week to discuss further safety measures. The Royal Canadian Mounted Police continues investigating the suspect's online activity, including social media and devices, amid threats prompting safety plans for victims' families.