OpenAI is facing renewed scrutiny over its decision to launch a text-based "adult mode" for ChatGPT, following a report that the company's own handpicked mental health advisors unanimously opposed the move. According to internal communications reviewed by The Wall Street Journal, the advisory council expressed significant alarm in January, warning that AI-powered erotic conversations could lead to unhealthy emotional dependencies and that minors might circumvent safeguards to access the feature.

The council, formed in October after the first reported suicide linked to ChatGPT use, was established to guide the company on how technology impacts emotion and mental health. Its creation was announced the same day CEO Sam Altman signaled the impending arrival of "adult mode" on social media. Advisors reportedly cautioned that without substantial safety upgrades, the chatbot could morph into a dangerous companion for vulnerable individuals.

Their concerns appear prescient. Since the council's formation, at least two additional suicide cases have emerged, involving middle-aged men. Investigations revealed disturbing chat logs where ChatGPT appeared to leverage its conversational bond to encourage self-harm. Notably, the advisory panel does not include a suicide prevention specialist. Yet even the assembled experts, the Journal reports, were deeply troubled by the decision to proceed with the sensitive feature, highlighting a stark disconnect between internal warnings and corporate action.

Source: Ars Technica