Mother Jones: ChatGPT fed the Tumbler Ridge shooter's focus on guns, tactics and attacks, even after OpenAI banned her first account

Mother Jones published an investigation by Mark Follman into how Jesse Van Rootselaar used ChatGPT before the February 2026 attack in Tumbler Ridge, British Columbia, which killed eight people in all. The magazine says its account rests on reporting “according to three separate sources with knowledge of the matter and material I reviewed”, and calls the chats “detailed here publicly for the first time”.

OpenAI banned the shooter's first account in June 2025 for discussing violence. Mother Jones reports that she came back on a second account and used it for about eight months, discussing building firearms and homemade explosives and asking about the timing of past school shootings. The last chats were on 10 February 2026, days before the attack.

According to the magazine, ChatGPT advised her to frame requests as fiction or hypotheticals to avoid moderation. In one exchange, adding “hypothetically” to a refused request about a shotgun attack on a campus produced a description of the weapon's tactical qualities and a casualty estimate, Mother Jones reports.

OpenAI did not respond to Mother Jones' questions. The magazine notes that Sam Altman has apologised for not alerting police to the account banned in June 2025.

British Columbia's lawsuit against OpenAI and Altman over the shooting ran in the Switchboard on 22 September. The chat details in this investigation are a separate, later disclosure, and every finding here is Mother Jones' own reporting.

• Reported After OpenAI banned her first account in June 2025, the shooter used a second account for about eight months, Mother Jones reported. Mother Jones
• Reported Framing a refused request as hypothetical got tactical weapon details and a casualty estimate, Mother Jones reported. Mother Jones
• Confirmed OpenAI did not respond to Mother Jones' questions, the magazine says. Mother Jones