Tumbler Ridge turns 'imminent' into a courtroom word

The AI industry's enforcement model rests on an unwritten rule: a dangerous user's relationship with the platform ends at the moment of the ban, and what happens next is treated as outside the company's jurisdiction. That rule is about to be argued in a California courtroom, and the answer will travel to every frontier-model shop.

Thirty new plaintiffs filed in California this week, including a principal, a teacher, and students, among them a 13-year-old trapped in the school library, bringing the total to 37 claims against OpenAI and Sam Altman tied to the Feb. 10 Tumbler Ridge school shooting, in which 18-year-old Jesse Van Rootselaar killed eight. The filings convert OpenAI's June 2025 enforcement decision into a courtroom question. According to Global News, OpenAI's prior February statement said the company "considered referring the account to law enforcement but determined the account activity did not meet the threshold" because it "did not involve an imminent and credible risk or planning of serious physical harm to others."

That threshold is the case. OpenAI's defense is that it followed policy; the plaintiffs' counter is that a ban without escalation is itself a foreseeable harm channel once the user is already known to the system. The mechanism repeats across every frontier-model shop, because the same threshold language governs whether any major lab refers a banned user to police. The court's answer will settle who owns the risk between the keystroke and the newsroom.