OpenAI debated calling police about suspected Canadian shooter’s chats

An 18-year-old who allegedly killed eight individuals in a mass taking pictures in Tumbler Ridge, Canada, reportedly used OpenAI’s ChatGPT in ways in which alarmed the corporate’s employees.

Jesse Van Rootselaar’s chats describing gun violence had been flagged by instruments that monitor the corporate’s LLM for misuse and banned in June 2025.

Employees on the firm debated whether or not or to not attain out to Canadian legislation enforcement over the conduct however in the end didn’t, in accordance to the Wall Street Journal. An OpenAI spokesperson mentioned Van Rootselaar’s exercise didn’t meet the factors for reporting to legislation enforcement; the corporate reached out to Canadian authorities after the incident.

“Our ideas are with everybody affected by the Tumbler Ridge tragedy,” an OpenAI spokesperson mentioned in an announcement. “We proactively reached out to the Royal Canadian Mounted Police with info on the person and their use of ChatGPT, and we’ll proceed to assist their investigation.”

ChatGPT transcripts weren’t the one regarding a part of Van Rootselaar’s digital footprint. She apparently created a sport on Roblox, the world simulation platform frequented by youngsters, which simulated a mass taking pictures at a mall. She additionally posted about weapons on Reddit.

Van Rootselaar’s instability was additionally identified to native police, who had been known as to her household’s dwelling after she began a fireplace whereas beneath the affect of unspecified medication.

LLM chatbots constructed by OpenAI and its rivals have been accused of triggering psychological breakdowns in customers who lose grip on actuality whereas conversing with digital fashions. A number of lawsuits have been filed that cite chat transcripts that encourage individuals to commit suicide or supply help in doing so.

If you’re in a disaster or having ideas of suicide, name or textual content 988 to achieve the 988 Suicide and Disaster Lifeline.

This submit has been up to date with remark from OpenAI.