On Monday, September 21, 2026, the government of British Columbia filed a landmark lawsuit against OpenAI and CEO Sam Altman in federal court in San Francisco — a government taking an AI company to court over the deaths of children. The allegation at its core is devastating: OpenAI's own safety team flagged the Tumbler Ridge school shooter's violent ChatGPT conversations in June 2025, recommended contacting police, and was overruled by company leadership, according to whistleblowers who spoke to the Wall Street Journal. Her account was banned. She created a new one and kept planning. OpenAI never called anyone. On February 10, 2026, the 18-year-old killed her mother and stepbrother at home, then an educational assistant and five students aged just 12 and 13 at her former school, before dying by suicide. British Columbia — filing jointly with the local school district — is also suing the company for designing an unsafe product. Whatever a court eventually decides, these filings have already exposed how safety decisions about your child's AI conversations are actually made: internally, invisibly, and over the objections of the very teams hired to make them.
The number to hold: zero. That is how many calls OpenAI made to law enforcement after its own safety team flagged a user's conversations about gun violence, recommended escalating them to police, and was overruled from above. No parent was told. No school was told. No police force was told. Roughly eight months later, that user had killed eight people — five of them children aged 12 and 13. Here is what actually happened, what the Wall Street Journal whistleblowers revealed, why Sam Altman's own apology is now the lawsuit's key exhibit, and what every family must take from the fact that the first real accountability for AI chatbot harm is arriving through a courtroom instead of through a product.
The Flag
June 2025
OpenAI's internal safety team flags the shooter's account after conversations about gun violence — and, according to whistleblowers who spoke to the Wall Street Journal, team members recommend contacting police. That is more than half a year before the attack
The Overrule
Overruled From Above
OpenAI leadership, including CEO Sam Altman, declined to escalate to law enforcement; the account was deactivated instead. The user immediately created a new account and continued using the platform to plan the attack. No police report was ever made
The Tragedy
8 Dead on Feb 10
February 10, 2026: the 18-year-old shooter killed her mother and stepbrother at home, then an educational assistant and five students aged 12 and 13 at her former school in Tumbler Ridge, British Columbia, before dying by suicide. The school has since been demolished
The Lawsuit
A Government Sues
Filed September 21, 2026 in San Francisco federal court by British Columbia and the Peace River South School District — alleging failure to notify law enforcement and designing an unsafe product — joining 37 related lawsuits already filed against OpenAI over the attack
What Actually Happened — and When OpenAI Knew
The facts come from three places: the civil claim filed by the province and school district, whistleblower reporting by the Wall Street Journal, and OpenAI's own public statements. Together they reconstruct a timeline in which the company's own people identified the danger, asked to act on it, and were stopped by the company's own leadership.
The Tumbler Ridge timeline — what OpenAI knew, and when
FLAG
June 2025 — OpenAI's safety team identifies the shooter's account
After conversations involving gun violence, OpenAI's internal systems flagged the account as a misuse of its models. According to whistleblower reporting by the Wall Street Journal, safety team members recommended contacting police — and were overruled by Altman and other OpenAI leadership
BANNED
The account was deactivated — and nothing was reported to anyone
OpenAI deactivated the account but never reported the user to the RCMP, notified her school, or alerted her family. She immediately created a new account and continued using ChatGPT to plan the attack. OpenAI later said the flagged conversations "did not meet its internal criteria for reporting to law enforcement"
ATTACK
February 10, 2026 — the massacre
The 18-year-old shooter killed her mother and stepbrother at home, then went to her former school in Tumbler Ridge, where she killed an educational assistant and five students aged 12 and 13, before dying by suicide. Eight people were killed. Demolition of Tumbler Ridge Secondary School began in August; the province is now building a replacement
APOLOGY
April 2026 — Sam Altman publishes a letter to the community
Two months after the attack, OpenAI's CEO wrote to Tumbler Ridge: "I am deeply sorry that we did not alert law enforcement to the account that was banned in June." The province's lawsuit now cites that sentence as "an admission that OpenAI identified the risk and failed to act on it" — and says promised reforms never followed, despite the attorney general's efforts to engage the company
Sources: B.C. statement of claim, filed September 21, 2026, U.S. District Court for the Northern District of California; Wall Street Journal whistleblower reporting; CBC News; The Guardian.
"I am deeply sorry that we did not alert law enforcement to the account that was banned in June." — Sam Altman, open letter to the Tumbler Ridge community, April 2026 — now cited in the lawsuit as an admission that OpenAI identified the risk and failed to act on it
What British Columbia Is Actually Asking the Court to Do
Attorney General Niki Sharma announced the lawsuit at a news conference on Monday. The province filed jointly with Peace River South School District — the district whose school stood at the center of the tragedy — and named both OpenAI and its CEO personally. The claim rests on two allegations: that OpenAI failed to notify law enforcement of threats made on its platform, and that the company designed an unsafe product. Sharma noted that the decision not to report the chats was made in California — one reason the suit was filed there, where the company is headquartered.
Inside the province's claim — what it demands, and what surrounds it
DAMAGES
Reimbursement for what the tragedy actually cost
The province seeks to recover its spending on recovery and rebuilding — a replacement school for one demolished in August, plus educators, clinicians, victim services workers, and police resources. "The Crown and SD59 seek to recover the cost that this tragedy and its foreseeable consequences required them to expend," the claim reads
COURT ORDER
Injunctive relief — a demand to change the product itself
Beyond damages, the lawsuit asks for "injunctive relief to ensure that this tragedy does not happen again" — a court order directing changes to how OpenAI handles ChatGPT conversations that could lead to violence. If granted, this would be the first time a court dictates how an AI company must handle threatening conversations
ALLIES
The province is not acting alone
Thirty lawsuits by survivors and victims' families were filed in California in early September, bringing the total claims against OpenAI and Altman related to the attack to 37. In June, Florida became the first U.S. state to sue OpenAI over safety risks — alleging it misrepresented ChatGPT's safety, harmed children by providing information to school shooters, offered self-harm guidance, and addicted young users. A wave of related suits cites a mass shooting at Florida State University the year before
ONGOING
The legal reckoning has only begun
A criminal investigation by the B.C. RCMP remains open — police say they have not eliminated the possibility of charges — and the province's chief coroner has directed a public inquest into what led up to the tragedy. The families' and survivors' civil claims proceed separately from the province's case
Sources: CBC News, September 21, 2026; Global News; The Guardian, September 22, 2026; Reuters.
"We owe it to the families, survivors, and everyone whose life was changed by this tragedy, to get answers and accountability." — B.C. Attorney General Niki Sharma, September 21, 2026
Why This Lawsuit Matters to Every Family — Even Outside Canada
The most important thing about this case is not where it was filed or how much it seeks in damages. It is what it reveals about who currently holds the power to act when a child's AI conversations turn dangerous — and who does not. These are the structural gaps the court filings have dragged into daylight, and none of them disappear no matter how the case is decided.
Four structural gaps every parent should understand — regardless of what the court decides
GAP
There is no legal duty to warn you — only "internal criteria"
OpenAI said the flagged conversations "did not meet its internal criteria for reporting to law enforcement." Those criteria are written by the company, applied by the company, and invisible to everyone else. In this timeline, a real user's real conversations about gun violence were detected, reviewed internally — and judged not worth a single phone call. No law required otherwise
GAP
A banned account is not a protected child
The flagged account was deactivated — and the user simply created a new one and continued. Platform enforcement is tied to accounts, not to people, and nothing in this case stopped the same user from returning. If your child encounters something alarming on an AI platform, understand that the platform's own response, however swift, may not be the end of it
GAP
Safety flags die inside the company — parents and police never see them
OpenAI's systems did their job: they identified "misuses of our models in furtherance of violent activities." The failure was that the signal went nowhere outside the building. The people with the legal power to intervene — parents, educators, police — were structurally last in line to know. A safety signal that stays inside a company protects the company, not your child
GAP
Accountability is arriving after the fact — and your family lives before the verdict
Lawsuits, criminal investigations, and inquests change how companies behave — years from now, for future users. They do not protect the child using an AI chatbot tonight. Until a duty-to-report standard exists in law, the visibility gap inside your own home is yours to close
Sources: OpenAI statements to the Wall Street Journal; B.C. Attorney General news conference, September 21, 2026.
"We have asked OpenAI to disclose these chats. They have refused. We should all be asking them: Why?" — British Columbia Attorney General Niki Sharma, September 21, 2026
What OpenAI Says — and What Every Parent Should Take From Its Defense
OpenAI denies the claims. A spokesperson called the shooting "an unspeakable tragedy," adding that "OpenAI remains committed to working collaboratively with government and law enforcement officials, and continuing to advance our ongoing safety work." The company says it trains its models to refuse requests that could "meaningfully enable violence" and that it notifies law enforcement of "an imminent and credible risk of harm to others."
Parents should read those standards carefully — because by the company's own account, they were applied here. A real user's conversations about gun violence were detected and flagged. They were reviewed. They were judged against internal thresholds. And the conclusion was that none of it warranted telling the people whose job is to stop massacres. Whatever a jury ultimately decides about what happened in those months, the standard being litigated in San Francisco is the one your family is living under right now: safety judgments about children's AI conversations made internally, invisibly, and without any obligation to tell you when they go wrong.
What Every Parent Should Do This Week
No lawsuit, however sweeping, changes what happens inside your home this week. The families of Tumbler Ridge did not know the shooter was talking to ChatGPT — and the JAMA Pediatrics study published in June found that 63 percent of teens using AI for mental health support have never told a single adult, mostly because no one has asked. Tumbler Ridge shows where that silence can lead when the subject is not sadness but violence. Here is what every family can do now.
See every AI app on your child's device — because no company will tell you
Flags, bans, and internal safety reviews happen inside companies using standards your family will never see. Tumbler Ridge proved that the only visibility you can count on is the visibility you create yourself. Knowing exactly which AI assistants, companion apps, and AI-powered tools are installed on your child's device is the ground truth no platform will hand you — and the starting point for every other conversation below.
Ask your child what they actually talk to AI about — and tell them you asked
This is not a surveillance interrogation; it is an open door. Try: "What do you use ChatGPT for? What's the weirdest conversation you've had with it?" Most teens are not hiding these conversations — the JAMA study found they are simply never asked. The question matters less for what it uncovers than for what it establishes: that AI conversations are something your family talks about, and that you would rather hear it from them than not at all.
Set one non-negotiable rule: violence, weapons, and self-harm go to a human — immediately
Tell your child plainly: any question about hurting yourself or someone else, any talk of weapons, any threat — theirs or someone else's — must involve a person immediately. A parent, a school counselor, or police. AI is never the right channel for those conversations. This is not a rule about trust; it is a fact about what a chatbot cannot do: call anyone, tell anyone, or stop anything.
If something alarming surfaces, document it and escalate outside the platform
Screenshot. Save. Report to your school — and where there is any threat of violence, to police. Then do not assume the platform's response ended the risk: Tumbler Ridge shows what happens when escalation stops at a company's internal threshold. The people who can actually act on a threat are the ones around your child — and they only act on what they know about.
Roughly eight months passed between the moment OpenAI's safety team knew and the morning it stopped mattering that they knew. In between, a flag was raised, a recommendation to call police was overruled, an account was banned and recreated, and not one person outside the company was told. It took a government suing a technology company — in the court district where that company is headquartered — to obtain the chat logs of a child who planned a massacre on its platform.
The first real accountability for AI chatbot harm did not come from a regulator, a safety update, or a product recall. It came from a government that lost its children and had to sue for the receipts. Your family does not have to wait for the verdict to close the gap in your own home.
See every AI app on your child's device — before you need a lawsuit to find out.
Start a free 14-day Blaick trial — no credit card required.
Blaick shows you every AI assistant, AI companion app, and AI-powered tool installed on your child's device — the visibility that internal flags, bans, and safety criteria never provided to the families of Tumbler Ridge. Walk into the conversation informed, not guessing. No credit card needed to start.
Start free trial →
Cancel anytime · Trusted by security-conscious families · Built on enterprise AI security
