Edelson PC is filing 30 additional lawsuits against OpenAI this week in California court, expanding a legal campaign that began in April with seven complaints tied to the Tumbler Ridge school shooting. Those bringing the new claims are students, teachers and a principal. All were present when the attack happened and none were physically hurt.
The filings mark a legal escalation, not just a numerical one. Where the earlier suits alleged negligence, the fresh filings raise the charge to aiding and abetting. That is a materially harder claim: it turns on intent, and it usually has a rougher time surviving a motion to dismiss.
The underlying case concerns a February 10 shooting at Tumbler Ridge Secondary School in British Columbia, Canada, in which a teenager killed eight people, including family members, before dying by suicide. According to reporting by the Wall Street Journal, OpenAI staff had flagged the attacker’s ChatGPT conversations, and employees urged leadership to alert Canadian law enforcement. Leadership declined. The company deactivated the user’s account; she opened a new one shortly after.
OpenAI has defended that decision by pointing to an internal threshold: conduct must present an “imminent and credible risk” of serious physical harm before the company will contact police. Jason Kwon, OpenAI’s chief strategy officer, told TechCrunch in a statement that “that judgment is not infallible, but it is always rooted in looking out for this balance for people.”
The new complaints go further than the earlier ones by naming Chris Lehane, OpenAI’s chief global affairs officer, as the person the plaintiffs allege told staff to stand down. The complaints themselves acknowledge they have no direct evidence of his involvement, stating the claim is made “on information and belief,” a legal term for an assertion based on secondhand information rather than confirmed proof. One complaint alleges the threat-assessment team was placed under Lehane’s control and that the no-contact decision was “ratified by Sam Altman.” Lehane is not named as a defendant; Altman is, as in the original seven suits.
Kwon disputed the allegation directly. “It is absolutely false to say Chris Lehane was involved with our original referral decision, or that our investigators report to him in any way,” he said, adding that it is “completely untrue” that political or public-relations considerations factored into the decision. TechCrunch reported it could not independently confirm whether Lehane has authority over the threat-assessment team or was involved in this case.
Jay Edelson, who is leading the litigation, said his firm is holding back the full evidence for now. He told TechCrunch the claim leans partly on how OpenAI is structured internally, and on a wider look at Lehane’s role within it. He said Lehane and Altman will be “critical witnesses.”
To rebut OpenAI’s “imminence” defense, the new filings reach back to November 2025, when the company put its San Francisco offices into lockdown, alerted employees and called police over a threat said to come from an activist, all despite what OpenAI itself called “no indication of active threat activity.” The plaintiffs argue that when OpenAI’s own employees were the ones potentially at risk, the company acted before waiting for a threshold it later applied more strictly to an external user.
That contrast points to the actual product and legal question underneath the litigation: “imminent and credible risk” is not a regulatory standard set by any outside body. Every AI company currently defines that threshold for itself, decides internally who applies it, and answers to no external reviewer when it declines to escalate. Until a regulator or a court sets a floor for when a provider must contact law enforcement, that judgment call will keep sitting entirely inside the company being sued over it.
For any AI provider running its own threat-assessment process, this case is now a test of whether “we decided internally that the bar wasn’t met” survives discovery once the decision-makers are deposed.
Reporting by Rebecca Bellan for TechCrunch, published September 2, 2026.
