OpenAI drops plans to release an adult chatbot

OpenAI has “indefinitely” deserted plans to launch an a erotic chatbot for adults following considerations from staff and traders, the corporate confirmed to The Financial Times. Plans for such a function, first introduced in October 2025 for launch in December final yr, had already been delayed whereas firm debated whether or not to launch all of it. It is the second app OpenAI has determined to shelf this week, after saying on Tuesday that it was shutting down its Sora video generator.

The adult-oriented chatbot, reportedly known as “Citron mode,” is now on maintain with no deliberate launch date. The corporate reportedly had problem coaching fashions that beforehand prevented erotic content material and likewise eradicating unlawful habits like bestiality or incest, two individuals conversant in the matter instructed the FT.

Open AI mentioned that it needed to conduct long-term analysis on the consequences of erotic chats and person attachment to AI, including that there was presently not but sufficient “empirical proof” on the topic. The corporate additionally mentioned it needed to concentrate on its core productiveness instruments like coding assistants and drop “aspect quests” like Sora and the erotic chatbot.

The concept for grownup options got here after OpenAI introduced that it will add parental controls and automatic age detection features for ChatGPT. CEO Sam Altman mentioned again in October that the corporate had at all times been cautious about such points over considerations round unhealthy AI attachments, however felt comfy that it may “safely chill out the restrictions generally.”

Nonetheless, the grownup mode had reportedly induced concern amongst traders, significantly amid the controversy attributable to rival xAI’s Grok model that generated deepfake nudes of actual individuals and youngsters. Workers additionally fearful concerning the function, with one senior worker even leaving the corporate over the difficulty. “AI shouldn’t substitute your mates or your loved ones; you must have human connections,” he instructed the FT.

One other problem is OpenAI’s age-checking tech, launched following lawsuits from households who mentioned that ChatGPT harmed their kids. The tech reportedly has an error proper greater than ten p.c, which might nonetheless give numerous younger individuals entry to the tech. OpenAI mentioned that determine is within the business normal vary and that it’s persevering with to work on its accuracy.