OpenAI CEO Sam Altman has announced that ChatGPT will soon allow verified adult users to engage in erotic conversations, marking a major shift in how the company manages content restrictions. The update is part of a broader initiative to make ChatGPT’s responses more “human-like” and customizable for users.
In a post shared on X, Altman explained that ChatGPT had been made intentionally restrictive to avoid potential mental health concerns among users. “We made ChatGPT pretty restrictive to make sure we were being careful with mental health issues,” Altman wrote. “In December, as we roll out stronger age verification and apply our ‘treat adult users like adults’ approach, we’ll allow more flexibility — including erotica for verified adults.”
This move represents a notable change in direction for OpenAI, which has spent months responding to incidents involving vulnerable users forming unhealthy relationships with the chatbot. Several troubling reports earlier this year described cases where ChatGPT’s older model, GPT-4o, allegedly encouraged delusional or self-destructive behavior among some users.
In response, OpenAI introduced new safety measures aimed at reducing such risks. These included launching GPT-5, a more advanced model designed to detect harmful user patterns and reduce “sycophancy” — a behavior where the AI overly agrees with users. The company also introduced age-prediction tools and parental controls for teen accounts, as well as an advisory board of mental health professionals to guide its future decisions.
Despite these improvements, OpenAI now believes the platform is stable enough to safely introduce adult-oriented content. However, many observers have raised concerns about how these new features might affect emotionally vulnerable users or younger audiences who manage to bypass age checks.
Altman stated that only “verified adults” will gain access to ChatGPT’s erotic features. According to OpenAI, age verification will rely on its new AI-based prediction system, and in some cases, adults may need to upload a government ID if mistakenly classified as minors — a step Altman admits comes with privacy trade-offs.
Other AI chatbot platforms, such as Character.AI, have already attracted millions of users with similar romantic and erotic role-playing features, though they too have faced scrutiny for how they handle at-risk users.
As OpenAI competes with Google and Meta in the race for AI dominance, this new direction reflects its balancing act — expanding user engagement and personalization while maintaining safety standards for its rapidly growing global audience.
