OpenAI CEO Sam Altman has apologized to members of a Canadian group the place a mass taking pictures passed off earlier this yr for not flagging the ChatGPT account of the shooter to regulation enforcement.
“The ache your group has endured is unimaginable,” Altman wrote in a letter shared Friday on social media by the British Columbia Premier David Eby. “I’ve been pondering of you typically over the previous few months.”
Eight individuals had been killed within the Feb. 10 bloodbath within the small group of Tumbler Ridge in northeast British Columbia. Six individuals had been fatally shot when 18-year-old Jesse Van Rootselaar opened fireplace at Tumbler Ridge Secondary College, authorities mentioned, and the shooter’s mom and 11-year-old brother had been killed at a close-by residence. Van Rootselaar died of a self-inflicted gunshot wound, officers mentioned.
Altman wrote within the letter, dated Thursday, that Van Rootselaar’s ChatGPT account had been banned in June 2025 — about eight months previous to the taking pictures.
“I’m deeply sorry that we didn’t alert regulation enforcement to the account that was banned in June,” Altman mentioned.
In February, OpenAI instructed CBS Information that Van Rootselaar’s account had been flagged final yr by automated abuse detection instruments and human investigators that determine potential misuses of ChatGPT for violent actions. OpenAI mentioned the account was then banned for violating its utilization insurance policies.
OpenAI mentioned that the corporate had weighed whether or not to flag the account to regulation enforcement, however had decided on the time that it didn’t pose an imminent and credible threat of significant bodily hurt to others, failing to fulfill the edge for referral.
“Our ideas are with everybody affected by the Tumbler Ridge tragedy,” OpenAI mentioned in a press release to CBS Information in February following the taking pictures. “We proactively reached out to the Royal Canadian Mounted Police with data on the person and their use of ChatGPT, and we’ll proceed to help their investigation.”
OpenAI says that ChatGPT is educated to discourage real-world hurt, and is instructed to refuse to assist when it detects a bootleg intent. Customers that point out plans to hurt others are flagged to human reviewers who decide whether or not a case poses an imminent risk of bodily hurt and must be referred to regulation enforcement, in keeping with the corporate.
Altman wrote in his letter that OpenAI will stay targeted on preventative efforts “to assist guarantee one thing like this by no means occurs once more.”
“I need to specific my deepest condolences to the complete group,” Altman mentioned. “Nobody ought to ever should endure a tragedy like this.”
Earlier this week, Florida Legal professional Normal James Uthmeier introduced a prison investigation into OpenAI after reviewing messages between ChatGPT and a Florida State College pupil accused in an April 2025 campus taking pictures that killed two individuals and wounded a number of others.
Uthmeier mentioned his staff decided that ChatGPT supplied “important recommendation” to the alleged shooter. His workplace is issuing subpoenas to OpenAI requesting data of the corporate’s protocols for reporting doable crimes to regulation enforcement, and its dealing with of person threats.
Concerning the Florida taking pictures, an OpenAI spokesperson mentioned in a press release to CBS Information Tuesday that “after studying of the incident,” it “recognized a ChatGPT account believed to be related to the suspect and proactively shared this data with regulation enforcement.”
