Some like it bot ChatGPT promises AI-rotica is coming for verified adults (www.theregister.com)

OpenAI CEO Sam Altman announced plans to relax ChatGPT’s behavioral restrictions so the assistant can be “more human-like” again — including personalities that use emoji and “erotica for verified adults” once age‑gating rolls out more fully in December. Altman framed the move as possible because OpenAI has formed an Expert Council on Well‑Being and rolled updates to GPT‑5 Instant that better detect emotional distress and route sensitive conversations to models trained for safer responses. He said the company has “mitigated” mental‑health risks, though the council’s work is ongoing and OpenAI has not published detailed evidence of those mitigations. The announcement is significant because it marks a deliberate tradeoff between expressive, human‑like assistants and safety controls, with major unanswered technical and operational questions. Key implications include how OpenAI will implement robust age and identity verification, prevent creation or distribution of non‑consensual or underage sexual content, and audit models fine‑tuned or policy‑layered to generate erotic material. There are also copyright and likeness risks analogous to OpenAI’s Sora work, plus monetization incentives that could shape design choices. For the AI/ML community this highlights the need for transparent safety evaluations, verifiable content‑filtering mechanisms, and concrete governance around consent, verification, and downstream harms before expressive capabilities are widely released.