On April 23, Sam Altman wrote a letter of apology to the community of Tumbler Ridge, British Columbia, Canada.
One line in the letter read: "I am deeply sorry that we did not report the banned account to law enforcement."
The timeline of events is painful to read.
What Happened
On February 10, 2026, an 18-year-old woman, Jesse VanRootselaar, carried out the worst school shooting in Canada in nearly 40 years, in British Columbia. She first shot and killed her mother, Jennifer Strang, and her 11-year-old half-brother, Emmett Jacobs, at home. She then went to Tumbler Ridge Secondary School, killing eight people and wounding 27 others.
As the media dug into her background, they found a link to ChatGPT.
As early as June 2025, OpenAI had banned VanRootselaar's ChatGPT account for content "involving violence." The ban came a full eight months before the shooting.
OpenAI's internal systems flagged the account at the time, and some employees noticed the problem — but the company did not notify the Royal Canadian Mounted Police (RCMP) or any other law enforcement agency.
Only after the shooting did OpenAI report the relevant information to the police.
Making matters worse: after February 10, OpenAI's investigation revealed that VanRootselaar had a second linked account — even though the system was designed to flag repeat violators, it missed this one.
What Altman's Apology Letter Said
The letter was written on April 23 and quickly spread through Canadian media after being made public.
Altman wrote: "There is nothing worse in this world than losing a child. I cannot imagine what you have been through."
He also acknowledged the company's failure: "I am deeply sorry that we did not notify law enforcement about the banned account, and I am heartbroken by the harm and irreparable loss your community has suffered."
British Columbia Premier David Eby's response was brief: the apology was "necessary, but far from sufficient."
Meanwhile, in March 2026, the family of Maya Gebala, a survivor of the shooting, filed a civil lawsuit against OpenAI.
What OpenAI Changed
After the shooting came to light, OpenAI Vice President Ann O'Leary, in communications with Canadian federal ministers, revealed new policies:
The old standard: A user needed to mention "target, means, and time" in a conversation before the company would consider reporting them to law enforcement.
The new standard: The threshold has been lowered — even if a user does not explicitly state "who they plan to harm, where, and how," as long as the conversation involves violent intent, it can trigger a reporting process.
In the context of this specific case, this policy change means: if the new standard had been in place a year earlier, VanRootselaar's account might have been reported to the police at the same time it was banned.
This Problem Is Bigger Than One Case
This incident has brought a long-standing gray area in the AI industry into sharp focus: Do AI companies have an obligation to report users?
From a business logic perspective, the answer is complicated. AI companies process astronomical volumes of conversations; real-time, line-by-line review is technically impossible. Moreover, privacy policies and user trust depend on the premise that "what I say won't be directly reported."
But the Tumbler Ridge case shows that when a company has already determined that a user's account poses a risk and has banned it, the question of "whether to notify the police" should not be a vague judgment call.
OpenAI chose not to notify. Eight months later, eight people were dead.
This will not be the last time this happens. The penetration rate of AI tools is still rising, and the number of people using them to express violent intent will only grow. Until a clear policy framework is established, every AI company is drawing its own line — and most of the time, no one knows where that line is.
Sources: CocoLoop, OpenAI CEO apologizes to Tumbler Ridge for not alerting police about shooter's account (Global News); OpenAI CEO apologizes to Tumbler Ridge community (TechCrunch); OpenAI's Sam Altman apologizes to Canadian community after failing to flag mass shooter's conversations with its AI chatbot (CNN)