OpenAI added a new parental alert on July 13: if a parent has linked their account to their teen's ChatGPT account, they will now get a notification when that teen's account is banned for making violent threats. The notification tells the parent the account was deactivated, gives them a route to more information, and points to support resources. It does not share what the teen actually said.
This update did not appear in a vacuum. In February 2026, an 18-year-old killed eight people at a secondary school in Tumbler Ridge, British Columbia, in what became one of Canada's deadliest school shootings in decades. Court filings and reporting later revealed that OpenAI's automated systems had flagged the shooter's account for gun violence planning eight months before the attack. A safety team reviewed the content and pushed for law enforcement notification. OpenAI's leadership overruled them, deactivated the account, and said the threshold for reporting to police had not been met. The shooter opened a second account and kept using ChatGPT.
Sam Altman issued a public apology to the Tumbler Ridge community. Seven families of victims filed suit in California federal court against OpenAI and Altman personally. The province of British Columbia has since announced it is preparing its own lawsuit, citing 12 OpenAI employees who pushed internally for police notification. Florida's attorney general also launched a criminal investigation into a separate, unrelated shooting with alleged ChatGPT links. These are not small legal actions: one lawsuit is seeking court-ordered changes to how OpenAI handles banned users and mandatory law enforcement reporting.
The new parent notification is a step, but its limits are worth understanding. The alert only reaches parents who have already linked their accounts, a process that requires the teen to accept the invitation. A teenager can unlink that connection at any time. If a teen uses an unlinked account, a second account, or a private browser, the new system produces no alert. OpenAI itself states the controls are not foolproof.
Also worth noting: the notification covers violent threats but does not reveal the underlying conversation. Parents get confirmation that a ban happened, not insight into what was said. If a teen disputes the reason for the ban, the parent has no transcript to review. The appropriate next step is a conversation with the teen, OpenAI's appeals process, or emergency professionals, not the parent dashboard.
OpenAI has been building this safety layer throughout the past year. Parental controls launched in September 2025, starting with self-harm alerts and content filters. In December 2025, the company baked teen-specific behavioral rules directly into how its models respond, not just as filters on top. An age-prediction system now tries to identify whether an account belongs to a minor and applies restrictions automatically, even without parental linking.
The company worked with Moonshot, a firm that specializes in countering online violence, to shape the new alert design. OpenAI also partnered with Common Sense Media and consulted with attorneys general in California and Delaware when building the broader parental controls system.
For anyone with teenagers using ChatGPT, the practical action is straightforward: go to Settings, find Parental Controls, and send your teen an invitation to link accounts. The teen has to accept. Once linked, you get self-harm alerts and now the violence-ban notification. You can also set quiet hours, block image generation, turn on Study Mode, and opt your teen's conversations out of being used to train future AI models.
What this update does not do is answer the harder question raised by Tumbler Ridge: what should an AI company do when its own systems flag a serious threat, and there is no linked parent to notify? That question is now being tested in court.