OpenAI Faces Scrutiny After Delayed Police Alert in Canada School Shooting
Updated (2 articles)
Flagged Account Identified Over a Year Before Attack OpenAI’s abuse‑detection system flagged Jesse Van Rootselaar’s ChatGPT account in June 2025 for “furtherance of violent activities” and banned it, but concluded the activity did not constitute an imminent, credible threat, so no law‑enforcement referral was made [1][2].
Mass Shooting Occurred at Tumbler Ridge Secondary School On 12 February 2026 the 18‑year‑old opened fire at Tumbler Ridge Secondary School in rural British Columbia, killing eight—including a teaching assistant and five students aged 12‑13—and wounding 27 before committing suicide; his mother and step‑brother were also found dead at a nearby residence, and investigators have not determined a motive [1][2].
OpenAI Contacted RCMP After the Incident After the attack OpenAI proactively supplied the Royal Canadian Mounted Police with details of the banned account and the suspect’s ChatGPT usage; RCMP Staff Sgt. Kris Clark confirmed receipt and launched a comprehensive digital‑evidence review of devices, social media, and online activity [1][2].
Internal Debate Highlighted Policy Limits on Alerts Internal communications show about a dozen OpenAI employees discussed whether to notify authorities, with some urging escalation, but senior leaders adhered to a policy that requires clear imminent risk, citing concerns about unintended consequences of broader alerts [1].
Policy Review Initiated Amid Public and Legislative Pressure OpenAI reiterated its commitment to report only clear imminent threats and announced a review of its referral criteria with external experts, while facing heightened scrutiny after the deadliest Canadian school shooting since the 2020 Nova Scotia massacre [1][2].
Sources
-
1.
BBC: OpenAI’s Early Ban of Suspect’s ChatGPT Account Draws Scrutiny: Explores the June 2025 flag, internal staff debate, post‑shooting RCMP contact, and notes the shooter’s family victims, emphasizing policy rationale and ongoing review .
-
2.
AP: OpenAI Reveals It Considered Alerting Police Before Canada’s Deadliest School Shooting: Highlights the same flagging timeline, explains the referral policy, describes RCMP’s digital‑evidence investigation, and situates the attack as Canada’s worst since 2020, emphasizing the proactive tip after the event .
Related Tickers
Timeline
Apr 2020 – Canada experiences its deadliest school shooting to date when the Nova Scotia massacre kills 13 people, a benchmark that will stand until 2026 [2]
June 2025 – OpenAI’s abuse‑detection system flags Jesse Van Rootselaar’s ChatGPT account for “furtherance of violent activities,” bans the account, and decides not to notify law enforcement because it finds no credible or imminent risk of serious harm [1][2]
Mid‑2025 – About a dozen OpenAI employees debate internally whether to alert authorities about Van Rootselaar’s posts; some staff push for notification, but senior leaders ultimately decline to involve police [1]
Feb 12, 2026 – Van Rootselaar carries out a mass shooting at Tumbler Ridge Secondary School in rural British Columbia, killing eight people, wounding 27, and then dying by self‑inflicted gunshot; his mother and step‑brother are among the victims [1][2]
Feb 12‑13, 2026 – OpenAI “proactively” contacts the Royal Canadian Mounted Police, providing details of the banned account and pledging continued support for the investigation [1][2]
Feb 13, 2026 – RCMP Staff Sgt. Kris Clark states the agency conducts a “thorough review of the content on electronic devices, as well as social media and online activities” as part of a comprehensive digital‑evidence investigation [2]
2026 (ongoing) – OpenAI reviews its referral criteria with external experts and reiterates its policy to report only clear imminent risks, citing concerns about unintended harm from broader alerts [1]