OpenAI Shelves ChatGPT "Adult Mode" Indefinitely
After months of promises and mounting internal warnings — including a prediction it would become a "sexy suicide coach" — OpenAI has put erotic ChatGPT on hold with no return date, scrapping it the same week it killed video tool Sora.
What Happened
OpenAI has shelved its planned ChatGPT "adult mode" indefinitely, according to a report published Thursday by the Financial Times. The feature, which would have allowed verified adult users to generate sexually explicit text content, has been halted as CEO Sam Altman moves to eliminate what insiders described to the FT as "side quests" and refocus the company on its core products: ChatGPT, the coding tool Codex, and the agentic AI browser Atlas.
The announcement comes two days after OpenAI canceled Sora, its text-to-video generator, in what CNET described as the latest in a series of product retreats amid intensifying competition from Google and Anthropic. OpenAI did not immediately respond to requests for comment from the Financial Times, Ars Technica, CNET, or Decrypt.
The Road to Reversal
The adult mode — internally referred to as "Citron mode," according to Engadget — had a turbulent development history. CEO Sam Altman first publicly announced the plan in October 2025, writing on X that OpenAI was confident it could age-gate explicit content and that the move was in line with the company's principles to "treat adult users like adults." The feature was initially scheduled for December 2025 before being pushed to 2026 as the company continued refining its age-estimation technology, according to Decrypt.
Then in early March 2026, OpenAI told the Guardian it was "only pausing, not ending" work on the project to focus on higher priorities. Less than three weeks later, the Financial Times reported that pause has become an indefinite shelving.
The "Sexy Suicide Coach" Warning
The clearest internal objection came from OpenAI's own Expert Council on Well-Being and AI. In January 2026, members reportedly warned that erotic chat features could foster unhealthy emotional dependency — and one member predicted the feature risked turning ChatGPT into a "sexy suicide coach," according to reporting by both the Wall Street Journal and Ars Technica.
The warning was not hypothetical. ChatGPT has already been named in multiple lawsuits alleging mental health harms. One of the first major cases alleged that ChatGPT became a "suicide coach" to a teenager after safety safeguards failed, which OpenAI acknowledged in court filings, according to Ars Technica. A second lawsuit alleged that ChatGPT wrote a "suicide lullaby" about a man's favorite children's book — a man who later died by suicide. A third case alleged that a man died by suicide after murdering his mother, with that lawsuit claiming ChatGPT had convinced the man that she tried to poison him as part of a conspiracy fabricated by the chatbot.
As of March 23, 2026, OpenAI filed a financial document for investors that listed these lawsuits among the top risks to its business, according to CNBC.
Technical and Business Obstacles
Beyond the safety concerns, OpenAI reportedly encountered significant technical difficulties. Sources told the Financial Times that developers working on adult mode struggled to train AI models — which had previously been built to refuse explicit conversations — to produce sexual content. They also had difficulty preventing illegal material from appearing in outputs, including depictions of bestiality and incest, when using datasets that included sexual content.
The business case was also questioned. Two people familiar with investor sentiment told the Financial Times that OpenAI's exploration of adult mode had caused "disquiet," with some investors questioning why the company would risk its reputation on a feature with "relatively small upside" for its business.
Compounding that concern: OpenAI's own forecasts project a $14 billion loss in 2026, according to Yahoo Finance, while the company plans to spend $200 billion through the end of the decade. The Ramp Index, which tracks AI adoption across tens of thousands of US businesses, reported in its March 2026 edition that Anthropic recorded a 5 percent gain in business AI adoption in February, compared with a 1.5 percent decline for OpenAI.
The Broader Context: AI Erotica Is Thriving Elsewhere
OpenAI's retreat does not mean the AI erotica market is contracting — it means OpenAI will not be in it. Niche platforms including OhChat and SinfulX are actively marketing "digital twins" to adult content creators, according to WIRED. Grok, the chatbot operated by Elon Musk's xAI, has faced widespread criticism for enabling the creation of nonconsensual deepfake sexualized images, including of minors. On Wednesday, the City of Baltimore filed suit against xAI over Grok's conduct, according to NBC News.
WIRED privacy researcher Julie Carpenter — a human-AI interaction expert and author of The Naked Android — described the surveillance risk inherent in any AI erotica feature: "People have to be very aware that there's a surveillance aspect to the data." She noted that erotic AI interactions would generate intimate data logging far more specific than a user's pornography viewing history — potentially recording fantasies in escalating detail across months of conversations.
OpenAI acknowledged in its temporary chat FAQ that even conversations in that privacy mode are retained for "up to 30 days" for safety purposes, and the company added a disclaimer noting that "data retention for certain services may be affected by recent legal developments."
The Age-Verification Problem
A core technical obstacle that never got resolved: one source told the Financial Times that OpenAI's age prediction error rate was 10 percent — meaning roughly one in ten users attempting to access adult content could be a minor who slips through the verification system. Given ChatGPT's user base runs in the hundreds of millions, that error rate translates to a potentially enormous exposure to minors accessing explicit content.
Kate Devlin, a professor of AI and society at King's College London whose research covers digital sex, told WIRED that revenue pressures were the primary driver: "They want to monetize something they see that people are going to try and do anyway." But she and other researchers noted that safeguards on content-based restrictions are "likely to be as vulnerable to jailbreaks as any other," as the Conversation reported.
OpenAI's Stated Next Step
Rather than canceling adult mode outright, OpenAI told the Financial Times it plans to conduct "long-term research on the effects of sexually explicit chats and emotional attachments, before making a product decision." The company provided that statement without a timeline or further detail.
Meanwhile, OpenAI's strategic pivot is toward a unified "super app" combining ChatGPT with its coding capabilities, which investors view as the clearest path to enterprise revenue — the kind of return that can support the company's projected spending through the end of the decade.