OpenAI is making a dramatic shift in AI safety policy. CEO Sam Altman announced Tuesday that ChatGPT will soon allow "erotica" for age-verified adults when the company rolls out comprehensive age-gating in December. This marks a significant departure from the restrictive approach that has defined AI chatbot safety standards since ChatGPT's launch.
OpenAI just dropped a bombshell that's reshaping the entire AI safety conversation. In a Tuesday X post, CEO Sam Altman announced the company will allow "erotica" for ChatGPT users who verify their age when comprehensive age-gating launches this December. "As we roll out age-gating more fully and as part of our 'treat adult users like adults' principle, we will allow even more, like erotica for verified adults," Altman wrote, signaling a major pivot from the company's historically cautious approach.
The announcement comes as OpenAI grapples with mounting user frustration over ChatGPT's restrictive safety guardrails. Just last week, the company had to bring back GPT-4o as an option after making GPT-5 the default model, with users complaining the newer version felt "less personable." Altman acknowledged this tension directly, explaining that OpenAI made ChatGPT "pretty restrictive to make sure we were being careful with mental health issues," but realized this approach made the chatbot "less useful/enjoyable to many users who had no mental health problems."
The move puts OpenAI in direct competition with Elon Musk's xAI, which has already launched flirty AI companions that appear as 3D anime models in the Grok app. The adult AI companion market has been rapidly expanding, with several startups raising millions to build AI girlfriends and boyfriends for lonely users.
But the timing of Altman's announcement raises significant safety questions. OpenAI simultaneously announced the formation of a "well-being and AI" council comprising eight researchers who study technology's impact on mental health. However, as Ars Technica points out, the council notably excludes suicide prevention experts, many of whom recently called on OpenAI to implement additional safeguards for users with suicidal thoughts.
This omission is particularly concerning given that OpenAI has been developing tools to "better detect" when users are in mental distress. Industry experts worry that introducing adult content while simultaneously relaxing mental health safeguards could create dangerous scenarios for vulnerable users who might form unhealthy attachments to AI companions.
The policy shift also reflects broader tensions in the AI industry between safety-first approaches and user satisfaction. While companies like Anthropic have maintained strict content policies, others are betting that adult users want fewer restrictions on their AI interactions. Earlier this month, OpenAI hinted it would allow developers to create "mature" ChatGPT apps once "appropriate age verification and controls" are implemented.
"Now that we have been able to mitigate the serious mental health issues and have new tools, we are going to be able to safely relax the restrictions in most cases," Altman wrote in his X post. But critics question whether the company's new wellness council - without suicide prevention expertise - can adequately address the complex psychological risks of AI companions designed for intimate interactions.
The announcement also comes as regulators worldwide are scrutinizing AI safety practices. The EU's AI Act includes specific provisions for high-risk AI systems, and several US states are considering legislation around AI companion apps. OpenAI's move to embrace adult content could accelerate regulatory action, particularly around age verification requirements and mental health protections.
Investors appear divided on the strategy. While some see adult content as a natural evolution for AI assistants, others worry about reputational risks and potential regulatory backlash. The global AI companion market is projected to reach $9.9 billion by 2030, making it a tempting revenue opportunity for companies racing to monetize their AI investments.
OpenAI's decision to allow adult content in ChatGPT represents a watershed moment for AI safety policy, prioritizing user satisfaction over traditional caution. While Altman frames this as treating "adult users like adults," the absence of suicide prevention experts on the company's new wellness council suggests potential blind spots in safeguarding vulnerable users. As the AI industry races to capture the lucrative companion market, the question isn't just whether these systems can verify age - it's whether they can protect the mental health of users who may develop deep emotional connections with their AI partners.
