OpenAI apologizes over Tumbler Ridge
- OpenAI chief executive Sam Altman apologized on April 24 to Tumbler Ridge, British Columbia, for not alerting police after ChatGPT banned an account later linked to shooting suspect Jesse Van Rootselaar. - OpenAI said the account was banned in June 2025 over possible real-world violence, but not referred because it failed the company’s threshold for an “imminent and credible” threat to others. - The apology lands as OpenAI says newer rules would trigger a referral, and Connecticut lawmakers are advancing Senate Bill 5 before May 6. (engadget.com)
Sam Altman apologized to Tumbler Ridge after OpenAI failed to alert police about a banned ChatGPT account later linked to the town’s school shooting suspect. (engadget.com) (usnews.com) In a letter dated April 23 and published April 24, Altman wrote that he was “deeply sorry” OpenAI did not alert law enforcement about the account banned in June 2025. The account was later tied to 18-year-old Jesse Van Rootselaar. (tumblerridgelines.com) (usnews.com) Police say Van Rootselaar killed eight people at a school in Tumbler Ridge on February 10, 2026, before taking her own life. CBC reported most of the dead were children. (usnews.com) (cbc.ca) OpenAI said it detected the first account through automated tools and human review for misuse tied to violent activity. The company banned the account, but said the June 2025 chats did not meet its internal standard for referral to police. (cbc.ca) That standard required signs of an “imminent and credible risk” of serious physical harm. OpenAI’s explanation drew anger from British Columbia Premier David Eby and federal Artificial Intelligence Minister Evan Solomon. (cbc.ca) Eby said Altman’s apology was “necessary” but “grossly insufficient for the devastation done to the families of Tumbler Ridge.” Altman said he had spoken with Eby and Mayor Darryl Krakowka before making the letter public. (engadget.com) (dailyhive.com) (tumblerridgelines.com) The case grew wider in February when OpenAI said the suspect had used a second ChatGPT account after the first one was banned. The company said it found that second account only after Royal Canadian Mounted Police released Van Rootselaar’s name. (politico.com) (cbc.ca) OpenAI then said it had updated its rules and would now notify authorities when it sees “imminent and credible” threats even without a fully specified target, means, and timing. Ann O’Leary, the company’s vice president of global policy, said the June 2025 account would be referred under the newer approach. (engadget.com) (politico.com) The policy debate has now moved into state legislatures in the United States. In Connecticut, Senate Bill 5 passed the Senate on April 21 and heads to the House before the General Assembly’s May 6 adjournment. (ctmirror.org) (cga.ct.gov) The Connecticut bill would require chatbot operators to make reasonable efforts to detect suicidal ideation or indicators of self-harm and to respond with appropriate resources. It also includes broader rules on employment tools, synthetic media, and frontier model safeguards. (senatedems.ct.gov) (legiscan.com) Altman’s letter promised OpenAI would work with governments to prevent a repeat. Tumbler Ridge is still measuring that promise against a warning that came months too early to ignore and too late to stop the killings. (tumblerridgelines.com) (engadget.com)
