OpenAI Says New Security Updates Would Have Sent Tumbler Ridge Shooters’ Posts to Police

Artificial intelligence firm OpenAI says its recently improved safety protocols will now identify online behaviours similar to those exhibited by the Tumbler Ridge shooter on its platform last year, and promptly report them to the authorities.

OpenAI has faced criticism and calls for reform after it was disclosed that the company flagged and banned a ChatGPT account in June 2025 belonging to the individual responsible for a mass shooting in Tumbler Ridge, B.C., earlier this month that left eight people dead and 27 injured. OpenAI owns and operates ChatGPT.

The account was not brought to the attention of the RCMP until after the Feb. 10 shooting because the company did not identify “credible” or “imminent” planning for potential real-world violence when it banned the account last summer, OpenAI said in a Feb. 26 letter to Artificial Intelligence Minister Evan Solomon. 

“With the benefit of our continued learnings, under our enhanced law enforcement referral protocol, we would refer the account banned in June 2025 to law enforcement if it were discovered today,” OpenAI vice-president for global policy Ann O’Leary said in the letter to Solomon.

OpenAI said it has already beefed up its protocols and company representatives committed to making further security improvements after meeting with federal ministers this week.

Two meetings took place in Ottawa at Solomon’s request and also included Justice Minister Sean Fraser, Public Safety Minister Gary Anandasangaree, and Culture Minister Marc Miller.

“Our conversation with you this week underscored that Canadians expect continued concrete action and we heard that message loud and clear,” O’Leary said in the letter to Solomon. “In our meeting, you and the other Ministers stressed that no community should have to face this tragedy. We agree.”

The letter said the shooter’s initial ChatGPT account was terminated last summer for violating the platform’s usage policies. It noted OpenAI’s automated systems first flagged the account, after which it was escalated for human review to determine whether any policies had been breached and whether the matter warranted referral to law enforcement.

“Based on what we could see at that time the account was banned in June 2025, we did not identify credible and imminent planning that met our threshold to refer the matter to law enforcement,” O’Leary said.

The letter also confirmed that the company discovered a second account belonging to the shooter after Jesse Van Rootselaar’s name was released to the public by the RCMP.

She said Van Rootselaar managed to circumvent the mechanisms designed to prevent prohibited users from establishing new accounts. She said the secondary account was disclosed to law enforcement upon its identification.

The letter said OpenAI is dedicated to improving its detection systems to more effectively thwart efforts to bypass its safeguards and to “prioritize identifying the highest-risk offenders.”

Calls for Reform

Details of Van Rootselaar’s internet activity began to emerge in the days after the attack, including that the teen had created an online game simulating a shooting massacre in a shopping mall and had been banned from ChatGPT last year.

U.S. newspaper The Wall Street Journal reported that while the violent posts on Van Rootselaar’s ChatGPT account last summer alarmed staff, the firm decided against contacting police.

O’Leary said in her letter the company would update its protocols about contacting police so that “when conversations cross the line into an imminent and credible risk,” such content is immediately reported. She also said OpenAI would develop a direct point of contact with Canadian law enforcement.

British Columbia Premier David Eby said the changes OpenAI is now making to its protocols are a “cold comfort” for the families of Tumbler Ridge.

Van Rootselaar’s first victims were the 18-year-old’s mother and 11-year-old half-brother. The armed teen then went to Tumbler Ridge Secondary School and shot a teacher’s aide and five student ranging in age from 12 to 13. Police say Van Rootselaar then died from self-inflicted gunshot wounds. 

Eby told reporters on Feb. 26 that OpenAI CEO Sam Altman has agreed to meet with him to discuss the matter.

“I think it’s important that Mr. Altman hear about how his team’s decision not to bring this information forward has resulted in the devastation that I witnessed first-hand in Tumbler Ridge,” said Eby. The premier did not say who would attend the meeting or when it would take place.

Eby said AI firms should not be trusted to determine their own reporting thresholds, and emphasized the need for a national standard that includes a minimum reporting threshold.

“We need all companies operating on the same threshold across the country, and that will be our message to the federal government,” he said, adding that his attorney general has sent a letter to the federal government to offer B.C.’s assistance in crafting online harms legislation, including a national standard for AI.

Solomon has not commented on Eby’s call for a set of federal rules to govern AI reporting, but said in a media statement this week that  “all options are on the table” as the government develops a “suite of measures” to address online harms and other digital policy issues.

Solomon said he and his fellow ministers had expressed their “disappointment” to OpenAI during their Feb. 24 meeting about its decision not to warn law enforcement. He also said he was disappointed that the company failed to come to the table with “concrete solutions” so Canadians could feel comfortable that this kind of tragedy would not recur.

A spokesperson for Solomon said his office is “reviewing OpenAI’s letter carefully and will have more to say in the coming days.”

The Canadian Press contributed to this report.