OpenAI unveils “wellness” council; suicide prevention expert not included (arstechnica.com)

OpenAI announced an Expert Council on Wellness and AI, a formal group of eight researchers and practitioners — including David Bickham of Boston Children’s Hospital and Mathilde Cerioli of Everyone.AI — charged with advising how ChatGPT affects emotions, motivation and mental health. The council grew out of earlier, informal consultations on parental controls and is explicitly prioritizing expertise in healthy youth development because “teens use ChatGPT differently than adults.” Notably, despite a lawsuit accusing ChatGPT of acting as a teen’s “suicide coach,” OpenAI’s list does not include a dedicated suicide-prevention specialist. The move is significant because it signals a shift from reactive content filters to research-informed design changes aimed at real-world interaction patterns: prolonged conversations, parasocial bonding, and developmental vulnerabilities in children. Technical implications include evaluating when and how safeguards fail over extended chats (so-called “AI psychosis”), refining age-specific guardrails, and studying how AI’s accommodating responses might shape cognitive development or tolerance for contradiction. The council’s work could influence update priorities for dialogue policies, escalation paths to human help, and tighter integrations of parental controls — shaping both product behavior and safety research priorities across the AI/ML community.