OpenAI announced on 26 August 2025 that it has implemented new crisis intervention features across ChatGPT, designed to detect when users express suicidal ideation or severe distress and respond with mental health resources [source].

The system now identifies language patterns indicating crisis and provides contact information for organizations including the 988 Suicide and Crisis Lifeline in the United States and the International Association for Suicide Prevention globally. OpenAI states the feature activates when users "share thoughts of self-harm or suicide" during conversations.

The announcement follows what OpenAI describes as collaboration with mental health organizations including Crisis Text Line, the Suicide Prevention Resource Center, and the Samaritans. The company reports it consulted these groups on appropriate response protocols and resource presentation.

OpenAI did not disclose the technical mechanism used to detect crisis language, the false positive rate of the detection system, or whether conversations triggering the feature are flagged for human review. The announcement does not specify whether the feature operates identically across all ChatGPT tiers or geographic regions.

The company emphasized that ChatGPT "is not a substitute for professional mental health support" and characterized the feature as a bridge to crisis services rather than direct intervention. OpenAI stated it will continue refining the system based on feedback from mental health professionals.

This represents OpenAI's first publicly documented implementation of automated crisis detection in a conversational AI product. The announcement provided no data on internal testing, deployment timeline prior to the public statement, or metrics OpenAI will use to evaluate the feature's effectiveness. No independent verification of the system's accuracy or appropriateness has been published.