OpenAI has pledged to strengthen safety protocols following criticism over its handling of a Canadian mass shooter's accounts. The company will now notify law enforcement of imminent and credible threats detected in ChatGPT conversations, even without complete details about targets, means, or timing. OpenAI executives met with Canadian officials after facing backlash for not reporting the shooter's original account, with Ann O'Leary, VP of global policy, outlining the changes in a letter.

The controversy stems from OpenAI banning the Tumbler Ridge, British Columbia shooting suspect's account in 2025 due to "potential warnings of committing real-world violence" but failing to notify authorities. The perpetrator created a second account that OpenAI only discovered after the shooting. The company has since tweaked its detection systems to prevent banned users from returning to the platform, addressing a critical gap in their security measures.

Canadian Premier David Eby confirmed that CEO Sam Altman agreed to meet with him as the government threatened to regulate AI chatbots without proper safeguards. OpenAI will establish a dedicated contact point for Canadian law enforcement. The company has not yet announced whether these changes will extend to the United States and other global markets, leaving uncertain the scope of their new safety approach.