ChatGPT-maker OpenAI stated Thursday the shooter in one of many worst faculty shootings in Canada’s historical past received round a ban on her problematic use of the service by having a second account.
The revelation got here because the San Francisco tech firm outlined in a letter to Canada’s authorities some “rapid steps” it was taking in response to the killings, and that if these had been in place on the time, police would have been knowledgeable of the exercise on the account.
OpenAI’s vice chairman for world coverage, Ann O’Leary, stated the corporate solely found the second account after Jesse Van Rootselaar’s title was introduced by the Royal Canadian Mounted Police, who stated Van Rootselaar killed eight folks after which herself in Tumbler Ridge, British Columbia, on February 10.
She stated the shooter one way or the other evaded techniques to forestall banned customers from creating new accounts, and Van Rootselaar’s second account was shared with regulation enforcement upon its discovery.
The letter stated OpenAI is dedicated to strengthening its detection techniques to raised stop makes an attempt to evade its safeguards and “prioritize figuring out the best threat offenders.”
The shooter’s first ChatGPT account was shut down in June 2025, the letter stated, after a violation of its utilization coverage. The letter stated OpenAI’s automated system detected the account, and it was then despatched to human assessment to find out whether or not its insurance policies have been violated and whether or not the account warranted referral to regulation enforcement.
“Based mostly on what we may see at the moment the account was banned in June 2025, we didn’t establish credible and imminent planning that met our threshold to refer the matter to regulation enforcement,” O’Leary stated.
Chatting with reporters on Thursday, British Columbia Premier David Eby stated Sam Altman, the CEO of OpenAI, has agreed to satisfy with him.
Eby stated his authorities was informed by OpenAI that adjustments to the thresholds to its protocols would have resulted in police being knowledgeable about Van Rootselaar’s ChatGPT exercise, had they been in place earlier than the killings.
However this was “chilly consolation” for the households of Tumbler Ridge, he stated.
In her letter, O’Leary additionally stated the agency will strengthen protocols about contacting police “when conversations cross the road into an imminent and credible threat.”
“With the advantage of our continued learnings, underneath our enhanced regulation enforcement referral protocol, we’d refer the account banned in June 2025 to regulation enforcement if it have been found at present,” she stated.
She stated OpenAI will develop a direct level of contact with Canadian regulation enforcement.
“The occasions in Tumbler Ridge are an unspeakable tragedy, and our hearts stay with the victims, their households, and your entire group,” O’Leary stated within the letter.
O’Leary thanked Canada’s Synthetic Intelligence Minister Evan Solomon for convening a gathering Tuesday to debate how you can assist stop related tragedies sooner or later.
“In our assembly, you and the opposite Ministers burdened that no group ought to must face this tragedy,” O’Leary stated. “We agree.”
Solomon referred to as OpenAI representatives to Ottawa to elucidate its security procedures and decision-making processes.
Solomon stated “all choices are on the desk” as the federal government develops a “suite of measures” to deal with on-line harms and different digital coverage points.
The RCMP stated Van Rootselaar first killed her mom and stepbrother on the household residence earlier than attacking the close by faculty. Van Rootselaar had a historical past of psychological well being contacts with police.
The motive for the taking pictures stays unclear.
The assault was Canada’s deadliest rampage since 2020, when a gunman in Nova Scotia killed 13 folks and set fires that left one other 9 useless.
Printed – February 27, 2026 10:03 am IST