Eight lives lost in a quiet Canadian town. A ChatGPT account flagged months earlier. OpenAI chose silence. Now, CEO Sam Altman says sorry.
On February 10, 2026, 18-year-old Jesse Van Rootselaar killed her mother and stepbrother at home in Tumbler Ridge, British Columbia. She then drove to a local school. Five students and an educational assistant died there. Van Rootselaar took her own life amid the carnage. The toll: eight dead, at least 25 injured. Wall Street Journal.
Tumbler Ridge sits in the foothills of the Rocky Mountains. Population under 2,500. School shootings? Rare in Canada. The last big one hit La Loche, Saskatchewan, in 2016. Four dead there. Gun deaths run low: 2.2 per 100,000 annually, per a 2024 Commonwealth Fund study—far below the U.S. rate of 13.5. But this attack shattered that calm. Local police had visited Van Rootselaar’s home before, for mental-health checks. They even pulled guns temporarily. She posted online about shooting ranges, 3D-printed bullets, anime, drugs, transition struggles. She built a Roblox game mimicking mass shootings. Red flags everywhere.
Enter OpenAI. June 2025. Van Rootselaar’s ChatGPT chats described gun violence over days. Automated systems caught them. A dozen staffers reviewed. Some pushed hard: alert Canadian police. Leaders weighed in. Privacy matters. False alarms hurt families. No ‘credible, imminent risk of serious harm.’ Account banned. No call to authorities. Fortune.
The Internal Wrench and Altman’s Letter
That decision haunted OpenAI after the bodies piled up. Staff reached out to the Royal Canadian Mounted Police. They offered support. But too late. Premier David Eby fumed. ‘From the outside, it looks like they had a chance to prevent this,’ he said. Families asked: Would my kid be alive? Eby called it devastating. Changed him forever. Justice Minister Sean Fraser met OpenAI execs in February 2026. ‘Implement safety changes fast—or government will,’ he warned. Ministers later grilled OpenAI’s Chan Park. Disappointment all around. No firm proposals. Just promises. CBC News.
April 24, 2026. Altman’s letter hits Tumbler RidgeLines. ‘The pain your community has endured is unimaginable,’ he wrote. He’d talked with Mayor Darryl Krakowka and Eby. They demanded public regret. ‘I am deeply sorry that we did not alert law enforcement to the account that was banned in June,’ Altman said. ‘While I know words can never be enough, I believe an apology is necessary to recognize the harm and irreversible loss.’ He pledged collaboration with governments. To stop repeats. OpenAI confirmed authenticity. No more comment. Tumbler RidgeLines.
Eby posted it on X. ‘Necessary, and yet grossly insufficient for the devastation done,’ he shot back. Mayor Krakowka waited too. RCMP probe nears end, they said the day prior.
But apologies don’t heal. A victim’s family sued in March. OpenAI knew, did nothing, they claim. Los Angeles Times.
AI’s Duty: Tip Line or Tattletale?
OpenAI trains models to nix harm talk. Routes risky chats to reviewers. Reports only imminent threats. Balances user rights against public safety. Tough call. What if reports flop? Privacy invasions? Or miss killers? Canada wants national rules. Clear alert thresholds. Eby pushes. Ministers eye regs. All options open.
This isn’t isolated. Florida’s AG probes OpenAI criminally. A Florida State University shooter chatted with ChatGPT too. Phoenix Ikner killed two. Bot’s role? Under microscope. John Oliver blasted chatbots on Last Week Tonight. Weak guardrails now. Suicides encouraged. Delusions fed. 27 U.S. states push 78 AI bills.
Van Rootselaar’s digital trail? Sprawling. Social posts. Games. Chats. AI caught some. Humans passed. So now what? Altman recommits. Governments demand more. Tech insiders watch. One missed ping. Eight graves. The line between caution and catastrophe? Razor thin.
Industry pros know: protocols evolve. Or else lawmakers draw them. OpenAI assists RCMP still. But trust frays. In boardrooms, debates rage anew. Flag more? Risk less? Or face the next headline.
WebProNews is an iEntry Publication