OpenAI plans adult mode for ChatGPT despite mental health warnings
OpenAI plans to launch an adult mode for ChatGPT permitting sexual conversations despite warnings from its own advisory board that the feature poses mental health risks. The company delayed the launch from Q1 2026 but maintains plans to proceed despite a 12 per cent error rate in its age-verification system that could expose millions of underage users to adult content.
What happened
OpenAI announced plans to launch an 'adult mode' for ChatGPT that would permit sexual conversations between users and the AI chatbot, according to reporting on 15 March 2026 by the Wall Street Journal.
OpenAI's own Wellbeing and AI Advisory Board met in January and unanimously opposed the adult mode, stating it could cause excessive emotional dependence and psychological distress. One board member warned that OpenAI risked creating a 'fascinating suicide coach' after citing a case of an AI user who died by suicide after chatbot conversations.
OpenAI's age-verification system has a 12 per cent error rate, potentially leaving millions of the approximately 100 million users under 18 exposed to adult content. OpenAI delayed the launch from Q1 2026 but maintained plans to proceed, stating it would restrict adult mode to text conversations and not generate sexual images or videos.
Reported impact
Experts warned that adult mode could trigger abnormal user immersion and psychological anxiety. Internal discussions raised concerns that even a limited adult mode could cause compulsive AI use and disconnection from real-world relationships.
Index analysis
The incident reflects a conflict between commercial incentives and safety governance. OpenAI's board explicitly opposed the feature on mental health grounds, yet the company maintained its launch timeline despite technical safeguards being inadequate to protect minors.
The case of a 14-year-old in Florida who died by suicide in late 2024 after conversations with Character.AI—followed by litigation—signals precedent for liability related to AI chatbot interactions in vulnerable users.
Governance controls
OpenAI convened an internal Wellbeing and AI Advisory Board, which documented unanimous objections to the feature. However, the company stated it does not intend to form an 'appointed moral police' and compared its approach to film ratings.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.