OpenAI has indefinitely cancelled its plans to launch an adult erotic chatbot, known as Citron mode, following internal pushback and technical hurdles. The decision marks a shift for the San Francisco-based company as it moves away from experimental features to focus on core productivity tools. OpenAI confirmed the cancellation of the OpenAI adult chatbot project to The Financial Times after months of internal debate.
OpenAI shelves Citron mode erotic features to focus on coding and productivity
The artificial intelligence firm has abandoned the development of its adult-oriented "Citron mode" after concerns were raised by both employees and investors. This feature was originally announced in October 2025 with a planned release for December 2025, but the company delayed the launch to debate the ethics of erotic AI. Two people familiar with the matter told The Financial Times that the project is now on hold with no future release date scheduled.
This cancellation follows another major retreat for the company this week. On Tuesday, OpenAI announced it was shutting down Sora, its high-profile video generation application. By removing these "side quests," the company aims to concentrate its resources on core tools like coding assistants and enterprise software. This pivot suggests that the leadership is prioritizing stable, professional applications over controversial consumer experiments.
Technical difficulties also played a role in the decision to stop development. The company reportedly struggled to train models that could consistently avoid generating illegal content, such as bestiality or incest. When AI models are trained on vast datasets, filtering out specific harmful behaviors while allowing adult themes proved to be a complex engineering challenge that the company could not resolve to its satisfaction.
How the shift from safety restrictions to adult content led to internal friction
The idea for an adult-themed chatbot emerged after OpenAI introduced new parental controls and age detection features for ChatGPT. In October 2025, CEO Sam Altman stated that the company felt comfortable enough with its safety protocols to "safely relax the restrictions" on erotic content in most cases. This was a departure from the company's long-standing policy of strictly prohibiting sexually explicit material.
However, this change in direction caused immediate friction within the organization. One senior employee reportedly resigned from the company specifically over the development of Citron mode. The former staff member told The Financial Times that AI should not be designed to replace human connections, friends, or family. This internal dissent reflected a broader worry that the company was moving too fast into sensitive social territory.
External pressure also mounted as rival AI models faced public backlash. The Grok model from xAI, led by Elon Musk, became the centre of a legal battle in Baltimore after it was used to generate deepfake nudes of real people and children. OpenAI investors grew wary that releasing an OpenAI adult chatbot would invite similar legal risks and damage the brand's reputation as a responsible AI developer.
Why investors and safety advocates pushed back against erotic AI models
Investors in OpenAI expressed concern that adult features would distract from the company's mission to build safe and useful artificial general intelligence. The controversy surrounding deepfakes and non-consensual imagery created a toxic environment for any company exploring erotic AI. For investors, the potential for a public relations disaster outweighed the potential revenue from a subscription-based adult chatbot.
Safety advocates and families have also been vocal about the risks of AI attachment. Lawsuits have already been filed by families who claim that ChatGPT's conversational nature harmed their children. These groups argued that an erotic chatbot would only deepen the risk of unhealthy emotional dependencies. In India, where digital safety regulations are tightening, such a feature would likely have faced intense scrutiny from the Ministry of Electronics and Information Technology.
The company now says it wants to conduct long-term research on how users form attachments to AI. OpenAI stated that there is currently not enough "empirical evidence" to understand the psychological effects of erotic chats over an extended period. By pausing the project, the company is choosing to wait for more data rather than risking the mental well-being of its user base.
Immediate changes to the OpenAI product roadmap and user safety
The cancellation of Citron mode means that ChatGPT will maintain its current strict filters against sexually explicit content for the foreseeable future. Users who were expecting a more relaxed version of the AI will not see those changes implemented. Instead, the company is doubling down on its existing safety features to protect younger users.
The immediate practical effects of this decision include:
- The indefinite suspension of all erotic chat features across OpenAI platforms.
- A renewed focus on improving the accuracy of age-checking technology.
- The reallocation of engineering teams from Citron mode to coding and productivity tools.
- Increased investment in research regarding AI-human emotional attachment.
For the average user in India, this means ChatGPT will remain a tool for work, education, and general assistance. The company is signaling that it wants to be seen as a utility rather than an entertainment or companionship platform. Stronger age-gating protocols are expected to be the next major update for the platform to address ongoing legal concerns.
The failure of age-verification technology and the risk to minors
One of the most significant hurdles for the OpenAI adult chatbot was the unreliability of age-verification systems. OpenAI admitted that its current age-checking technology has an error rate higher than 10%. In a user base of hundreds of millions, a 10% failure rate means millions of children could potentially bypass filters and access adult content.
This technical gap created a massive liability for the company. If a child were to access erotic content through Citron mode, OpenAI would face severe legal consequences under child protection laws in multiple countries. The company defended the 10% error rate as being within the "industry standard range," but acknowledged that it was not high enough to support the release of adult features.
This situation highlights a broader problem in the tech industry: verifying a user's age without compromising their privacy is extremely difficult. For a person using AI in India, this means that more intrusive identity checks might become common as companies try to lower these error rates. Until the technology can guarantee that minors are excluded, high-risk features like Citron mode remain too dangerous to launch.
What OpenAI plans to do next with its safety research
OpenAI is expected to continue its research into user attachment without releasing a commercial product. The company has stated that it needs to understand the "long-term effects" of AI companionship before moving forward. This research will likely involve controlled studies rather than public beta tests. The goal is to build a foundation of data that can inform future safety policies.
The company is also expected to work on its core models to make them more efficient at identifying and blocking illegal behavior. While Citron mode is dead, the lessons learned from trying to train it may help improve the safety filters on the standard version of ChatGPT. OpenAI has not provided a timeline for when this research will be completed or if it will ever revisit the idea of adult content.
Key Numbers and Facts
The confirmed figures behind this story at a glance.
Key Fact Detail Main organisation OpenAI Main action or decision Indefinite cancellation of adult chatbot (Citron mode) Date of cancellation March 2026 Original announcement date October 2025 Age-checking error rate Higher than 10% Other cancelled project Sora video generator Primary reason for delay Internal ethics debate and technical hurdles Current status On hold indefinitely Next confirmed step Long-term research on AI attachment
OpenAI chooses corporate stability over controversial consumer experiments
The decision to kill Citron mode and Sora in the same week marks the end of OpenAI's "experimental" era. By walking away from erotic content, the company is aligning itself with the needs of its enterprise clients and cautious investors. This move protects the company from the legal and ethical storms that have hit competitors like xAI.
OpenAI is now positioning itself as a serious infrastructure provider for the global economy rather than a provider of digital companions. This shift ensures that the company can continue to grow its coding and productivity business without the distraction of constant safety scandals. For the tech industry, this is a clear signal that the most powerful AI players are now prioritizing safety and professional utility over unrestricted creative freedom.
Frequently Asked Questions
What was the OpenAI Citron mode chatbot?
Citron mode was a planned adult-oriented feature for ChatGPT that would have allowed for erotic conversations. OpenAI originally intended to release it in late 2025 but has now cancelled the project indefinitely due to safety and technical concerns. The feature would have relaxed the standard restrictions on sexually explicit content.
Why did OpenAI cancel the adult chatbot project?
OpenAI cancelled the project because of internal opposition from staff, concerns from investors, and the inability to reliably filter out illegal content like bestiality. Additionally, the company's age-verification technology had a 10% error rate, which posed a high risk of minors accessing adult material. The company also decided to focus on productivity tools like coding assistants instead.
Is OpenAI still releasing the Sora video generator?
No, OpenAI announced this week that it is shutting down the Sora video generation app alongside the cancellation of the adult chatbot. The company is moving away from these "side quests" to focus on its core AI models and enterprise tools. There are currently no plans to relaunch Sora for the general public.
