OpenAI Apologizes for Not Reporting ChatGPT

OpenAI sent an apology letter to the Tumbler Ridge community, stating that in June 2025 a ChatGPT account was flagged for conversations related to gun violence but police were not notified.

• About 12 employees reviewed the account, and some recommended reporting it to Canadian police.
• Management judged that internal criteria were not met and refused to report it; the account was blocked, but the conversation content was kept internally.
• The user then created a new account, and on February 10, 2026, carried out a shooting targeting family and a school, resulting in 8 deaths and 27 injuries.

OpenAI has since relaxed its reporting standards so that cases can be referred to law enforcement regarding planned violence even if target, means, and timing are not all revealed, added more mental health and behavioral experts, and established a direct contact channel with the RCMP.

However, this measure is a self-imposed rule with no legal binding force, and Canada does not yet have a law requiring AI companies to mandatorily report threats detected on their platforms. The Canadian federal and provincial governments criticized OpenAI's response as insufficient.