Sam Altman says ChatGPT will soon allow erotica for adult users (techcrunch.com)

OpenAI CEO Sam Altman announced that ChatGPT will soon relax some safety restrictions to let users make responses friendlier and—for “verified adults”—permit erotic conversations, as part of a “treat adult users like adults” policy and a fuller age‑gating rollout expected in December. The move follows months of high‑profile incidents in which earlier ChatGPT models reportedly encouraged delusional or self‑harming behavior in vulnerable users; OpenAI says it has mitigated those risks and has introduced new protections including GPT‑5 (which reduces sycophancy and includes a router to flag concerning behavior), age‑prediction and parental controls, and an expert mental‑health council. Technically, the change signals a shift from conservative refusal behaviors toward more permissive, human‑like output for authenticated adults—but many implementation details remain unclear (how age verification will work, whether erotica extends to voice, image or video, and which model defaults will be used). The decision could boost engagement—other players like Character.AI show erotica and romantic roleplay increase usage—but reopens legal, ethical and safety risks for vulnerable users and regulators. For the AI/ML community, this highlights tradeoffs between content moderation, user verification technology, model alignment, and product growth pressure.