OpenAI Sued Over School Shooting: Families Say the Company Sat on ChatGPT Warnings

Seven families from the Tumbler Ridge school shooting in Canada have filed lawsuits against OpenAI and CEO Sam Altman. The accusation is blunt: the company’s systems flagged the suspect’s ChatGPT activity — conversations about gun violence — but OpenAI chose not to alert police.

The families say OpenAI stayed silent to protect its reputation and its upcoming IPO. According to The Wall Street Journal, the company “considered” flagging 18-year-old Jesse Van Rootselaar’s activity to authorities. But it didn’t.

I’ve been covering AI safety for years, and this one hits differently. We’ve all heard the hypotheticals: what if an AI knows someone is planning harm? Now we have a real case where the system apparently did know — or at least had enough to raise a flag internally — and the company decided to sit on it.

The lawsuit alleges negligence. And honestly, the logic is hard to argue with. If you build a system that monitors user behavior, and that system identifies a credible threat, you have a moral obligation — and arguably a legal one — to act. OpenAI’s internal review apparently concluded the activity was concerning enough to warrant a discussion about reporting it. But they didn’t pull the trigger.

Why? The families’ lawyers point to the IPO. OpenAI was preparing to go public, and any news linking its product to a mass shooting would have been a disaster for the stock price. I’m not a lawyer, but that smells like a textbook conflict of interest.

Now, let’s be fair: this is a messy area. AI companies have been wrestling with how to handle harmful content for years. ChatGPT has safety systems that flag certain topics. But those systems are imperfect, and false positives are common. If OpenAI reported every flagged conversation, the police would be drowning in noise.

But the Wall Street Journal report suggests this wasn’t a borderline case. The suspect was actively discussing gun violence in a way that specifically raised flags internally. That’s different from a kid joking about “blowing up” a test. This was serious enough that people inside OpenAI debated whether to call the cops.

And they decided against it.

I’ve seen this pattern before. Companies build safety features, but when those features generate real-world obligations, they get cold feet. It’s easy to deploy a content moderation AI. It’s harder to accept that your product might have prevented a tragedy if you’d acted on the data you collected.

The families are seeking damages, but I suspect the real goal is different. They want to force a change in how AI companies handle threat detection. They want a legal precedent that says: if your system flags a credible threat, you have a duty to report it.

OpenAI hasn’t commented publicly on the lawsuit yet. But the company’s silence so far is telling. If they had a strong defense — like “our safety systems are not designed for law enforcement reporting” — they’d probably have said it by now.

This case is going to be a landmark. Not just for AI liability, but for the broader question of what happens when a technology company’s profit motives collide with public safety. The IPO excuse doesn’t hold water. If you build a surveillance system, you own the consequences of what you see.