Thirty new lawsuits filed against OpenAI are raising an urgent question about artificial intelligence that goes far beyond what children should ask a chatbot: What responsibility does an AI company have when its technology detects signs that someone may be planning real-world violence?

The new complaints, filed September 2 in federal court in California, stem from the February 10, 2026, mass shooting in Tumbler Ridge, British Columbia. An 18-year-old killed her mother and 11-year-old half-brother at home before going to Tumbler Ridge Secondary School, where five children and an educator were killed. Twenty-seven other people were injured. The shooter later died by suicide.

The 30 new plaintiffs include students, teachers and a school principal who survived the attack. Their cases follow seven lawsuits filed earlier this year by victims and families, bringing the number of lawsuits connected to the shooting to 37. At the center of the litigation is ChatGPT and what OpenAI knew months before the attack.

Did OpenAI Know the Tumbler Ridge Shooter Was Discussing Violence?

OpenAI has acknowledged that its safety systems flagged the shooter’s original ChatGPT account in June 2025, approximately eight months before the shooting.

According to OpenAI, automated systems detected activity that violated its violence policies. Human investigators reviewed the account, and OpenAI ultimately banned it.

The company did not notify Canadian law enforcement. That decision has become central to the lawsuits. The plaintiffs allege that the shooter’s conversations with ChatGPT contained sufficiently serious warning signs that authorities should have been contacted. The latest complaints go further, alleging that members of OpenAI’s safety operation recommended a law-enforcement referral but were overruled.

Those are allegations, not established findings of fact. OpenAI strongly disputes portions of the plaintiffs’ description of its internal decision-making. In particular, the company denies allegations that Chief Global Affairs Officer Chris Lehane or public-relations considerations influenced the decision not to contact authorities.

What isn’t disputed is that OpenAI identified concerning activity, banned the original account, and decided at the time that the situation did not meet its threshold for notifying law enforcement.

Why Didn’t OpenAI Contact Police?

This is where the case becomes particularly important for families. Technology companies have to make extraordinarily difficult decisions when users discuss violence. People may talk to AI about fictional stories, news events, historical violence, anger, intrusive thoughts or other subjects that don’t mean they intend to hurt anyone.

Automatically reporting every disturbing conversation to police would create enormous privacy and civil-liberties concerns. But failing to act on a genuine threat can have devastating consequences.

OpenAI says its standard is to determine whether a situation presents an imminent and credible risk of serious physical harm to others. The company has said the Tumbler Ridge shooter’s activity did not meet that threshold at the time it was reviewed.

OpenAI CEO Sam Altman has since apologized to the Tumbler Ridge community for the company’s failure to alert authorities about the account.

Could the Shooter Simply Create Another ChatGPT Account?

The original account was banned, but that didn’t permanently prevent the shooter from accessing ChatGPT. OpenAI says she later created another account. The company has said it was not aware of that second account until after the shooting.

That raises another important child-safety question: What should happen after a technology company identifies a user as potentially dangerous and removes that person’s account? Account bans can stop access to a particular account. They don’t necessarily prevent someone from creating another identity, using another email address, switching devices, or finding another service.

OpenAI says it has since strengthened its ability to detect people attempting to evade enforcement after an account has been banned.

What Do the New OpenAI Lawsuits Allege?

The new lawsuits make several serious claims that will have to be tested through the legal process. Among other allegations, plaintiffs argue that OpenAI failed to adequately respond to warning signs in the shooter’s ChatGPT conversations and should have contacted law enforcement. Some of the new complaints also accuse OpenAI of aiding and abetting the attack, an allegation that raises a significantly higher legal bar than negligence.

OpenAI denies that its employees placed political or public-relations considerations ahead of safety. Jason Kwon, OpenAI’s chief strategy officer, said the company’s investigators and legal team work with mental-health, behavioral and law-enforcement experts when developing standards for identifying potential threats.

The lawsuits could ultimately force courts to confront questions that existing technology law was not necessarily designed to answer.

When Should an AI Company Warn Law Enforcement?

This may be the most important question arising from Tumbler Ridge. AI chatbots occupy an unusual position in people’s lives. Unlike a search engine that receives a handful of disconnected searches, a chatbot can participate in lengthy conversations and potentially encounter patterns over time.

That creates possibilities for identifying risk, but also difficult questions about surveillance and privacy.

Should an AI company contact authorities when someone asks a disturbing question? Probably not based on that alone. What if the person repeatedly discusses killing specific people?

What if they ask for help planning an attack? What if the AI system detects a pattern suggesting the threat is becoming more specific?

And how certain should a private technology company have to be before potentially giving someone’s private conversations to law enforcement? The Tumbler Ridge litigation may help define where some of those boundaries lie.

Has OpenAI Changed Its Safety Policies?

Yes. In its current community safety policies, OpenAI says conversations involving potentially serious real-world harm can be escalated for additional investigation. The company says that when conversations indicate an imminent and credible risk of harm to others, it notifies law enforcement.

OpenAI also says its assessment doesn’t necessarily require a user to explicitly provide a target, weapon, and exact timing before a situation can be escalated. Mental-health and behavioral experts help inform difficult assessments. The company says it has also strengthened detection of people attempting to return after being banned. Those changes are significant, but the lawsuits will examine whether different decisions should have been made before the Tumbler Ridge shooting.

What Does This Mean for Parents Whose Children Use AI?

Parents shouldn’t conclude from this case that a chatbot automatically creates violent behavior or that every disturbing conversation with AI predicts real-world violence. The lawsuits involve specific allegations surrounding one horrific event, and courts have not yet determined OpenAI’s legal responsibility.

But the case does expose something families need to understand about AI. A chatbot can appear private, personal, and even confidential to the person talking to it. In reality, AI platforms may use automated safety systems and human reviewers to identify certain types of dangerous activity, and companies may sometimes take action based on those conversations.

Parents should also avoid treating AI safety systems as emergency services. If a child talks about harming themselves or someone else, makes a specific threat, appears to be planning violence, or reveals that another person may be in immediate danger, don’t assume an AI platform will recognize the danger and alert someone for you. Contact appropriate emergency services, school officials or qualified professionals directly when circumstances require immediate intervention.

The Bigger Question Is No Longer Just Whether AI Is Safe for Kids

For years, the conversation around children and artificial intelligence has focused on misinformation, homework, privacy, inappropriate content and whether children might become emotionally dependent on chatbots.

Tumbler Ridge introduces another question. What happens when an AI system sees warning signs before the people around a user do?

The lawsuits don’t answer that question yet. They are allegations that will now move through the legal system, and OpenAI will have the opportunity to contest them. But 37 lawsuits tied to one devastating school shooting make the underlying issue difficult to ignore.

As artificial intelligence becomes increasingly woven into children’s lives, schools and private conversations, society will have to decide not only what AI should refuse to say, but when an AI company has seen enough that someone outside the chat needs to know.

---