OpenAI Tightens ChatGPT Violence Safeguards With Teen Weapons Limits and Chat Cut-Offs

For accounts the company estimates belong to users under 18, ChatGPT will limit practical information about weapons and security systems "even when a question appears harmless," AI Weekly reported. This goes beyond an existing ban on helping teens get weapons, which applied when there were signs of violent intent. OpenAI said it has strengthened its age-prediction system, which uses behavioural signals to move users it believes are minors into the teen version of ChatGPT. Teens can still use the chatbot to help prevent or calm down violence, seek support and contact a trusted adult.

Parents will soon get a notification if a teen's account is paused for repeatedly discussing violence, the company said. In July, OpenAI had already expanded parental alerts to cover teen accounts deactivated for breaking its violence policies.

Rules for All Users

Some of the changes apply to every user, whatever their age or reason for asking. ChatGPT will give less detail when asked to reconstruct past mass attacks, including what weapons were used and how they were obtained, AI Weekly reported, even where that information is available elsewhere online. Automated safety systems can end a conversation immediately when a user appears to be fixating on violence, and repeated violations can lead to a temporary account pause.

OpenAI also said it made more than 200 referrals to law enforcement in the third quarter of 2026 involving potential threats of violence, according to AI Weekly.

The company plans to publish a Community Safety Report on violence prevention every six months. The first edition, due in early 2027, will cover July to December 2026 and include anonymised examples of violence-related harm on the platform along with data on how well the safeguards work.

Why It Matters

The measures come as AI developers face growing legal and political pressure over how chatbots handle users at risk. OpenAI faces several lawsuits from US families who allege ChatGPT contributed to teen suicides or violent crimes. The company first introduced parental controls and age prediction in 2025 and launched a dedicated teen version of ChatGPT in August 2026.

The changes also involve trade-offs that businesses and institutions using ChatGPT will notice. Age prediction based on behaviour can put adults into the restricted teen experience by mistake, and AI Weekly asked how researchers, journalists and historians with legitimate questions about weapons or past attacks will be treated. Automatic chat cut-offs and account pauses also give OpenAI's classifiers more say over whether a conversation can go on.

The regular safety report could become a template for disclosure. US chatbot makers are not currently required to tell law enforcement when their systems flag possible violence, and regulators in several jurisdictions are weighing how much transparency to demand on AI safety incidents.