OpenAI Confirms ChatGPT Conversation Reviews, Police Notification in Threat Cases

OpenAI has confirmed that conversations with ChatGPT containing violent threats or plans to harm others may be reviewed by human evaluators and potentially reported to law enforcement if deemed an imminent threat. The company uses automated detection systems combined with human review to identify and escalate such cases through specialized pipelines authorized to take actions including account bans. However, OpenAI explicitly states that conversations involving self-harm or suicide ideation are flagged internally for safety resources but are not shared with police to respect user privacy. This disclosure has raised concerns among users who expected their interactions to remain confidential, with critics worried about risks such as impersonation or misuse leading to wrongful police involvement. OpenAI emphasizes its commitment to user safety by directing those expressing suicidal intent to professional help hotlines, while balancing privacy and security considerations in its monitoring policies. The policy aims to prevent serious physical harm while maintaining transparency about the potential for police intervention in extreme cases.