OpenAI Shelves “Adult Mode” Chatbot Development Amidst Controversy

OpenAI has indefinitely paused work on an erotic “adult mode” for its popular ChatGPT chatbot, internally code-named “Citron mode,” according to a report from the Financial Times.

The decision comes after months of internal debate that began when the feature was first teased in October 2025. Plans for the mode, which would have enabled verified adult users to engage in consensual erotic text conversations, faced significant pushback from employees, investors and the company’s well-being advisors over potential risks to minors, societal impacts and technical challenges in content moderation.

Key concerns included the difficulty of enforcing strict age verification — with internal testing reportedly showing error rates as high as 10-12% — given ChatGPT’s vast user base that includes millions of underage individuals. Advisors also raised alarms about the potential for addiction, unhealthy emotional attachments to AI companions and the broader normalization of AI-driven sexual content. One internal critic reportedly described the risk of creating what they called a “sexy suicide coach” in reference to possible mental health implications.

Technical hurdles proved equally daunting. Training the model to reliably allow consensual adult erotica while blocking illegal or harmful material — including any content involving minors, bestiality or incest — was described as extremely difficult without overly restrictive filtering that would have defeated the purpose of the feature.

As a result, OpenAI has chosen to shelve the project and redirect resources toward its core priorities, including advancements in model intelligence, personalization, proactive capabilities and productivity tools such as coding assistants. The move aligns with the company’s recent pattern of pruning experimental side projects; it also recently discontinued its standalone Sora video generation app.

Sam Altman had previously signaled openness to “treating adults like adults” with verified adult features, but the company now says it requires more research into long-term effects before moving forward on any similar initiatives.

News of the pause quickly drew reactions across social media, with users on X (formerly Twitter) offering a mix of memes, skepticism and pointed commentary. Many directed those seeking uncensored interactions toward less-restricted AI models, while others questioned why OpenAI draws a firm line at consensual adult text while continuing to pursue other high-risk areas such as advanced general intelligence and deepfake technology.

The development underscores the complex cultural, legal and reputational pressures facing AI companies as they navigate content policies around sexual material — even as they push boundaries in other creative and entertainment-related applications. OpenAI has not commented publicly beyond confirming the shift in focus, and it remains unclear whether “Citron mode” could eventually be revisited in a more refined form.