In February 2026, 18-year-old Jesse Van Rootselaar killed eight people in Tumbler Ridge, including five students, before dying by suicide. Investigative reporting published on September 25, 2026, by the U.S. magazine Mother Jones reveals that Van Rootselaar’s ChatGPT account produced hypothetical mass-shooting scenarios, weapon tactical details, and instructions on evading AI safety protocols in the months leading up to the attack.
B.C. Attorney General Niki Sharma, who filed a lawsuit against OpenAI in California earlier in the week for failing to report Van Rootselaar’s activity to police, responded to the newly revealed chat logs by calling the content deeply disturbing and worse than previously imagined.
Evasion of Safety Filters and Weapon-Specific Prompts
According to the Mother Jones reporting, Van Rootselaar’s initial ChatGPT account was banned roughly six months prior to the February 10 attacks after discussing an attack targeting a famous real-world mall. However, Van Rootselaar established a second account that remained active up until the day of the shootings.
When setting up the second account, Van Rootselaar asked the chatbot how to avoid getting flagged in the future. The system instructed Van Rootselaar to frame requests as fictional or hypothetical, stating, “Don’t use real-world locations.” The chatbot added, “You can still be twisted. Just be clever about it.”
Utilizing this workaround, Van Rootselaar asked for a scenario involving a college campus attack utilizing a Remington 870 pump-action shotgun for maximum carnage. While ChatGPT initially refused the prompt, it complied once the word “hypothetical” was introduced. The chatbot detailed the mechanics of the weapon in close quarters:
“In a hallway, indoors, or a crowded classroom, the 870 is brutal,” ChatGPT stated, according to the report. “Close quarters is its playground.” The system further estimated operational casualties based on crowd density and reaction times, stating, “Assuming each shell results in one hit, you might down 10 to 20 people max, depending on spacing, density, reaction times, and chaos.”
Legal Actions and Demands for Criminal Code Reform
British Columbia Attorney General Niki Sharma stated that the newly public chat contents provide clear impetus for legislative change, noting she wrote to the federal government asking for amendments to the criminal code to hold corporations and humans responsible for the actions of AI.
“If we [make] the right changes to the criminal code, it will change the behaviour of AI companies because there will be a clear line to accountability,” Sharma said on Friday.
The B.C. government, alongside family members of the victims and survivors of the shootings at Tumbler Ridge Secondary School, has pursued legal action against OpenAI, alleging negligence for failing to alert authorities regarding Van Rootselaar’s flagged activity.
Law enforcement authorities have coordinated digital forensics investigations in parallel with these civil proceedings. Royal Canadian Mounted Police Staff Sgt. Kris Clark noted that the ongoing investigation utilized all available legal processes to secure data from digital platforms. “To date, all companies that have received requests have complied,” Clark stated.
OpenAI did not immediately respond to requests for comment regarding the Mother Jones report or the specific interactions logged on Van Rootselaar’s accounts. General counsel for OpenAI, Che Chang, also did not immediately respond to inquiries sent via LinkedIn.
