OpenAI has indefinitely paused plans to launch a chatbot capable of engaging in erotic conversations for adult users, Financial Times reports. The feature, dubbed ”Citron mode,” was first announced in October 2025 with a planned December release but has faced multiple delays before being halted altogether.
Sources say the main challenge lies in training models that were originally designed with strict filters blocking explicit content. Ensuring the AI would not generate illegal or harmful scenarios-such as forbidden behaviors-also proved difficult. As a result, the project is now on hold without a clear timetable for revival.
OpenAI has emphasized the need for long-term research into how erotic AI interactions affect users, including potential emotional attachments. The company believes there isn’t yet enough empirical evidence to roll out such features safely. Instead, OpenAI plans to focus on its core productivity tools-like coding assistants-and discontinue side projects such as the recently shuttered video generator, Sora.
The adult chatbot concept emerged after OpenAI announced developments in parental controls and age verification for ChatGPT. CEO Sam Altman has expressed caution about these features, citing risks of unhealthy dependency on AI, though he acknowledged that restrictions might loosen over time.
Investor skepticism heightened following the controversy around xAI’s Grok model, which became embroiled in generating deepfake content involving real people. Internally, OpenAI has seen disagreements too: according to Financial Times, a senior employee resigned, opposing the direction that AI might replace human relationships.
Another sticking point is the age verification system implemented after lawsuits from families who claimed ChatGPT negatively affected their children. Reportedly, this system has a failure rate exceeding 10%, potentially allowing minors access to restricted content. OpenAI maintains this is in line with industry standards and continues efforts to improve accuracy.
For context, while companies like Google and Meta experiment cautiously with adult-themed AI features under tight content moderation, OpenAI’s hesitation reflects broader concerns about ethical boundaries and user safety. The challenges around age verification, content control, and emotional effects underscore the complexities of deploying adult content in AI chatbots at scale.
Looking ahead, OpenAI’s move signals a cautious recalibration as the company balances innovation with ethical responsibility. Whether Citron mode-or a similar adult feature-returns will depend on breakthroughs in safe content filtering and reliable age verification. The debate around AI’s role in human intimacy and dependency is far from over, and OpenAI’s cautious stance may shape industry norms for years to come.