A disturbing revelation about a mass shooting in Canada after reports revealed that OpenAI had flagged the account of the alleged shooter and did not report him to the authorities.
The Straits Times reported that OpenAI flagged and banned 18-year-old Jesse Van Rootselaar’s ChatGPT account about eight months before he allegedly murdered eight people in and around a school in Tumbler Ridge, British Columbia.
The company’s internal systems detected troubling prompts Rootselaar used related to gun violence. An OpenAI spokesperson told reporters the company “proactively reached out to the Royal Canadian Mounted Police with information on the individual and their use of ChatGPT, and we’ll continue to support their investigation.”
Van Rootselaar’s chats described alarming scenarios involving gun violence, using the platform to plan out the alleged attack. The company’s system flagged his interactions in June 2025. This triggered an internal review and debate among about a dozen OpenAI workers about whether the company should alert Canadian law enforcement.
The Wall Street Journal reported that some employees urged management to contact police. However, the decision makers eventually concluded Van Rootselaar's conduct did not meet the threshold for an “imminent and credible” threat that would justify violating a user’s privacy. Instead, the company chose to ban the account for violating usage policies.
The company said it “considered referring the account to law enforcement at the time, but did not identify credible or imminent planning and determined it did not meet the threshold.”
The shootings started at Tumbler Ridge Secondary School on February 10 where the gunman killed six people and injured at least 25 others. The police found Van Rootselaar dead from a self-inflicted gunshot wound.
Investigators later discovered the bodies of the suspect’s 38-year-old mother and 11-year-old stepbrother. This marks one of the worst mass shootings Canada has experienced.
Canadian school shooting suspect had a history of mental health struggles and prior police welfare checks, and authorities say the attacker killed family members before carrying out a deadly school shooting that left multiple people dead and injured.pic.twitter.com/JXpD7h9Mb5
— Brandon Straka #WalkAway (@BrandonStraka) February 12, 2026
OpenAI claims its systems are trained to detect and discourage “the possible furthering of violent activities,” and that the company is now reviewing when this behavior should trigger referrals to law enforcement.
Canada’s federal minister responsible for artificial intelligence said the case raised “serious questions” about whether voluntary corporate standards are enough to keep people safe. He suggested he and other government officials might use this shooting to expand the government’s control over artificial intelligence companies. Ottawa could consider imposing rules to lay out when AI platforms are required to report potential threats to the authorities.
OpenAI spokesperson Kayla Wood said the company is attempting to balance user privacy with public safety. It is working with regulators and other experts to avoid “unintended consequences from overly broad referrals to law enforcement.”
Stories like this have proliferated since artificial intelligence became more mainstream, with ordinary consumers adopting the technology. Similar to social media platforms, AI companies will likely struggle to strike the balance between public safety and privacy rights, as Wood suggested.
The exact wording of the prompts Van Rootselaar used is not yet known, so it is difficult to judge whether OpenAI made the right decision in this case.