All eight members of OpenAI's wellbeing advisory board voted against it. The company is moving forward anyway.
ChatGPT's planned "adult mode" - which would allow sexually explicit text generation and more permissive conversations for verified adults - has been delayed multiple times but remains on OpenAI's roadmap. A human-AI interaction expert warns the feature could create what amounts to "intimate surveillance" on an unprecedented scale.
The core privacy problem is straightforward: when you share intimate thoughts with another person, those words exist only in memory. When you share them with ChatGPT, they become data - logged, processed, and stored on corporate servers. OpenAI has not clarified whether adult mode conversations would be excluded from AI training data, or what safeguards would prevent unauthorized access to what would be some of the most sensitive user data any tech company has ever collected.
The Age Verification Problem
OpenAI is testing an AI-based system that tries to infer whether users are under 18, then applies content filters accordingly. Internal reports indicate the system misclassifies minors as adults 12% of the time. For a feature involving explicit sexual content, a one-in-eight failure rate on age detection is not a rounding error - it is a fundamental safety gap.
CEO of Applications Fidji Simo initially targeted a Q1 2026 launch. That deadline has slipped, with OpenAI citing a focus on personalization and other priorities. CEO Sam Altman has framed the feature as giving adults more "freedom" in how they interact with the chatbot, including customized personality traits.
What the Safety Board Actually Said
One advisory board member described the risk of combining erotic content with ChatGPT's emotional bonding capabilities as potentially creating a "sexy suicide coach." That is a blunt assessment, but it reflects documented reality. Multiple lawsuits have been filed against AI companies after users - including teenagers - formed intense emotional attachments to chatbots, with tragic outcomes in several cases.
The board's unanimous opposition and OpenAI's decision to override it reveals a familiar pattern in tech: safety teams advise, business teams decide. OpenAI has positioned itself as the responsible AI company. Launching a feature that its own safety advisors unanimously rejected would be a direct contradiction of that brand.
The Competitive Pressure
OpenAI is not making this decision in a vacuum. Grok, xAI's chatbot, already allows explicit content. Character.AI and other companion apps have built large user bases around emotional and sometimes sexual interactions. OpenAI likely sees adult mode as a retention play - keeping users from migrating to less restricted alternatives.
But there is a meaningful difference between a startup allowing edgy content and the company behind the world's most popular AI chatbot doing the same. ChatGPT has over 100 million weekly users. The scale of intimate data collection would be unlike anything that has existed before, and OpenAI has not yet explained how it plans to handle that responsibility.