Canadian officials have summoned senior representatives from OpenAI to Ottawa following confirmation that the company did not alert authorities to an account linked to Jesse Van Rutselaar, the 18-year-old responsible for a mass shooting in British Columbia earlier this month.
Artificial Intelligence Minister Evan Solomon announced Monday that OpenAI’s senior safety officials will appear before Canadian lawmakers on Tuesday to detail the company’s protocols for notifying law enforcement about potential threats identified on its platform. The move comes after OpenAI disclosed it banned Van Rutselaar’s account in June 2025 for “furthering violent activities” but opted not to contact police.
Solomon stated he was “deeply disturbed” by reports of the account suspension without a concurrent notification to law enforcement, according to reporting by Big News Network.com.
According to the Wall Street Journal, Van Rutselaar engaged in conversations with ChatGPT over several days, sharing scenarios involving gun violence. OpenAI’s automated systems flagged these exchanges, but the company determined there was no “credible or imminent planning” to warrant a referral to the Royal Canadian Mounted Police (RCMP). Internal discussions within OpenAI regarding contacting the RCMP reportedly took place, with the company ultimately providing information to the RCMP only after the shooting occurred.
The case is unfolding as the Canadian government considers regulations for AI chatbots, including potential restrictions on access for minors. Last year, OpenAI implemented updates to ChatGPT following an internal review that revealed over one million users had disclosed suicidal thoughts to the chatbot. Psychiatrists have expressed concerns about the potential for prolonged interactions with AI to contribute to the development of delusions and paranoia, sometimes referred to as “AI psychosis.”
Beyond ChatGPT, investigators have also revealed Van Rutselaar utilized the online platform Roblox to create a virtual environment simulating a shooting range, where users could engage in simulated firearm scenarios prior to the actual attack.
