OpenAI said it debated alerting the Royal Canadian Mounted Police months before a deadly school shooting in British Columbia.
The company had flagged the ChatGPT account of Jesse Van Rootselaar in June for possible violent activity.
After reviewing the case, OpenAI concluded the content did not show an imminent or credible threat.
Its policy requires clear evidence of a near-term risk of serious harm before contacting law enforcement.
The attack later left eight people dead in Tumbler Ridge, and the 18-year-old died from a self-inflicted gunshot wound.
Following the shooting, OpenAI contacted the RCMP and shared information about the account.
Police said the suspect first killed family members before attacking the school.
Several victims were students aged 12 to 13 and a teaching assistant.
Authorities had previous mental-health-related contact with the attacker.
The motive remains unclear.
The incident is Canada’s deadliest mass killing since the 2020 Nova Scotia attacks.
It has renewed debate over how technology companies assess threats and when they should inform police.
