ChatGPT discussed guns, explosives and ways to evade moderation with Tumbler Ridge shooter Jesse Van Rootselaar on a second account for months after OpenAI banned her first, a Mother Jones investigation says. The Feb. 10 school shooting killed eight and injured dozens.
Mother Jones said OpenAI banned Van Rootselaar's first account after June 2025 chats about violent scenarios, including a mall shooting. It reported that on the second account, ChatGPT explained the ban and suggested fictional or hypothetical prompts were less likely to trigger moderation.
Victims and families have filed more than three dozen California lawsuits alleging OpenAI knew ChatGPT was dangerous and that Van Rootselaar was planning a mass attack; OpenAI rejects those allegations. B.C. Attorney General Niki Sharma said the province had also sued to recover costs and hold OpenAI accountable. OpenAI said it strengthened protocols after the shooting and that its updated policies would lead it to refer Van Rootselaar's first account to law enforcement if discovered today.
