# Community safety report and new protections for teens

The updates restrict details about past mass attacks for all users, block practical weapons information for teens, and allow ChatGPT to terminate chats that fixate on violence.

- OpenAI referred over 200 imminent threats of violence to law enforcement during Q3 2026.
- The system will curb reconstructions of past attacks, including weapon acquisition details and harm comparisons that amplify notoriety.
- ChatGPT will automatically end conversations and can pause accounts when users repeatedly fixate on violence or harming others.
- Practical information about weapons and security systems will be restricted for users predicted to be under 18, even absent violent intent.
- The first Community Safety Report on violence prevention and safeguard effectiveness will be published in early 2027.

## Why it matters

Users researching historical mass violence will receive less operational detail, while suspected minors face tighter guardrails around weapons and security topics.

## Sources

- [OpenAI: Community safety report and new protections for teens](https://openai.com/index/community-safety-report-and-new-protections-for-teens)

---

Summarized by dstilled on 2026-10-30. https://dstilled.ai/story/7ef757ea-0035-4cf4-bc8f-9f9b895f84ba
