Full Breakdown
OpenAI’s Missed Threat and the Tumbler Ridge Mass Shooting
4/27/2026, 10:07:38 PM
Core Event
In June 2025 OpenAI’s automated abuse detection flagged a ChatGPT account used by 18-year-old Jesse Van Rootselaar for describing gun-violence scenarios. A dozen staff urged police referral, but leadership applied a higher internal threshold and did not alert authorities. He later opened a second account and on February 10 2026 carried out a mass shooting in Tumbler Ridge, British Columbia, killing eight people and injuring 27 before committing suicide.
Background & Context
OpenAI’s safety team classified the flagged chats as an “imminent risk of serious harm.” Employees escalated the concern, yet a spokesperson later said leadership imposed a “higher threshold” for credible threats, preserving the logs but making no police contact. The Wall Street Journal first reported the internal dispute.
Timeline
June 2025 – OpenAI flags Van Rootselaar’s account; staff recommend police, leadership declines.
February 10 2026 – Van Rootselaar murders mother and step-brother, then kills eight and injures 27 at Tumbler Ridge Secondary School before suicide.
April 23 2026 – CEO Sam Altman issues a public apology.
Late April 2026 – OpenAI announces a lowered reporting threshold and a direct RCMP liaison.
Official Statements & Responses
CEO Sam Altman expressed deep sorrow, pledged cooperation with all government levels, and vowed OpenAI would work to prevent future tragedies. Vice-president Ann O’Leary announced a lowered reporting threshold, mental-health expert review, and a direct RCMP contact point, noting the 2025 conversation would now be reported. Premier David Eby called the apology “necessary” but “grossly insufficient.”
Criticism & Opposition
AI minister Evan Solomon warned that OpenAI’s voluntary steps “do not go far enough” to safeguard the public. Critics highlight the dissolution of OpenAI’s internal safety, superalignment, and AGI-readiness teams and the shift from nonprofit to for-profit status, which they say weakens oversight. Seven families have sued over alleged “suicide coaching,” and a civil suit in British Columbia alleges the model gave weapon-selection guidance.
Conflicting Reports & Gaps
Reports vary on the death toll: some cite eight victims plus the shooter, others list ten deaths including the shooter. The flagged ChatGPT excerpts remain undisclosed, preventing verification of alleged weapon-selection advice. No police referral was made, and officials have not clarified whether the RCMP would have acted. Canada lacks a statutory duty for AI firms to report threats.
Verbatim Quotes
- “I am deeply sorry that we did not alert law enforcement to the account that was banned in June,” — Sam Altman, CEO, OpenAI
- “I cannot imagine anything worse in the world than losing a child.” — Sam Altman, CEO, OpenAI
- “The pain your community has endured is unimaginable.” — Sam Altman, CEO, OpenAI
- “necessary” but “grossly insufficient for the devastation done to the families of Tumbler Ridge.” — David Eby, Premier of British Columbia
What’s Next
Canada’s joint task-force of Innovation, Science and Economic Development Canada and Public Safety Canada will issue safety-reporting recommendations by summer 2026. Lawmakers are reviewing Bill C-27 and Bill C-63 for possible mandatory AI threat-reporting.
