Archive · 2 Sep 2025
OpenAI May Share ChatGPT Chats with Police, Citing Safety
OpenAI, the company behind ChatGPT, might share your conversations with the police if they think someone is in danger.
If you talk about hurting yourself, the AI will give you numbers to call for help.
But, if you say you want to hurt someone else, OpenAI will have people look at the conversation and might call the police.
The company is trying to make sure its AI is safe, but it admits that the safety features don't always work perfectly in long conversations.
OpenAI is also working on more safety features, like helping users with extreme sleep problems or unsafe activities.
OpenAI may share ChatGPT conversations with police if they indicate a risk of harm.
Human moderators review chats expressing intent to harm others.
Suicidal intent leads to resources like 988 in the US.
Imminent threats result in police alerts and possible account bans.
Safety mechanisms can degrade in long or repeated conversations.
- Who
- OpenAI
- What
- May share ChatGPT conversations with police.
- When
- Recently
- Where
- N/A
- Why
- Due to potential safety risks, specifically the risk of serious physical harm to others.
- Issue
- OpenAI may share user chats with authorities.
- Trigger
- Risk of serious physical harm to others.
- Action
- Chats reviewed by human moderators.
- Result
- Police alerted for imminent threats.
- Suicidal Intent
- Users directed to professional resources.
- Account Ban
- Accounts involved in threats can be banned.