OpenAI said on 30 October 2026 that ChatGPT will give teenagers less practical information about weapons and security systems that could be misused, and announced a Community Safety Report on how it handles violence. Teens get the tighter treatment even when a question looks harmless and shows no sign of violent intent.
Some limits apply to every user. ChatGPT will no longer reconstruct past mass attacks, including the weapons and equipment used and how they were obtained, or draw detailed comparisons of the scale of harm that could amplify an attacker's notoriety. Helping anyone plan violence, and helping teens find weapons, was already barred; the new rule extends that ban to practical weapons and security-system information. Teens can still use ChatGPT to prevent or calm violence, seek support and reach a trusted adult, and anyone can ask for help understanding past events, their victims and warning signs.
Where recent chats show signs of violence, ChatGPT will answer more conservatively, and it applies the stricter limits whenever it predicts a user is under 18. Age prediction reads behavioural signals; someone wrongly placed in the under-18 experience can confirm adulthood through Persona, OpenAI's age verification partner. ChatGPT can also end a chat when a user seems fixated on violence, and after several ended chats OpenAI may pause the account for a time. It is working on notifying parents when that happens to a teen. The safeguards roll out over the next few weeks, according to Unite.AI.
Header image Unite.AI ran with its report on OpenAI's Community Safety Report and teen weapons limits
Image: Unite.AI
The first Community Safety Report is due in early 2027, covering the second half of 2026, and will be updated every six months. The report is meant to explain how OpenAI detects, investigates and responds to safety issues, with anonymised examples, new mitigations and supporting data. OpenAI says it made more than 200 law enforcement referrals in the third quarter of 2026 and argues that AI firms have a distinct role alongside police, families, mental-health professionals and policymakers, and says its referrals of imminent threats are helping disrupt plots.
About OpenAI
OpenAI's handling of minors has been pressed before. In August 2025 the parents of a 16-year-old who died by suicide filed Raine v. OpenAI, a wrongful-death suit against the company and Sam Altman alleging that ChatGPT contributed and that its safeguards for minors were inadequate.
On 16 September 2025, days after the FTC opened a probe into AI chatbot companions and teens, OpenAI said users it identified as under 18 would be sent to an age-appropriate version of ChatGPT and announced parental controls with account linking and blackout hours. It had banned the Tumbler Ridge shooting suspect's account in June 2025 after gun-violence scenarios, but staff who debated alerting police did not. After the shooting, Altman apologised in a letter in late April 2026, and OpenAI said it would loosen its criteria for referring accounts to authorities.
11
Works credited
27
Models indexed
Models on AIPopDB
Parental controls launched in 2025 and alert parents of linked teen accounts when a teen breaks ChatGPT policy on violent threats or acts; in July 2026 the alerts were widened to accounts deactivated for violence. ChatGPT for Teens launched on 18 August 2026, and age prediction began reaching consumer plans on 20 January 2026. Engadget notes that OpenAI faced a wrongful-death lawsuit in 2025 and that Sam Altman apologised this year for not alerting police to chat logs tied to the Tumbler Ridge shooting.
Sources · 2