OpenAI Kills X-Rated ChatGPT Plan After Investor Revolt and Safety Panic

OpenAI has officially paused, “indefinitely,” any work on an adults-only ChatGPT mode that would have allowed sexually explicit conversations with verified users. The decision came Wednesday after a Financial Times report detailed heated internal dissent and a cold shoulder from investors who feared reputational fallout. According to The Verge, the move mirrors the company’s earlier shelving of Sora video generation—both side projects shoved aside to keep the core product roadmap intact.

The reversal is total: code has been mothballed, beta invites cancelled, and the internal Slack channel renamed to “archive-adult.” No new timeline has been offered, and sources tell Ars Technica the feature is now considered “radioactive” inside the company.

Money talked louder than product ambition. The FT reports that several late-stage funding leads explicitly warned OpenAI’s CFO that any whiff of pornographic AI could trigger ESG blacklists and complicate IPO plans. Venture firms already jittery about regulatory headlines saw the erotic mode as an easy line item to cut.

One partner at a top-5 VC fund told Reuters they’d “rather back a boring SaaS tool than explain to LPs why we financed a sexbot.” The optics problem was compounded by OpenAI’s nonprofit board structure, which gives outside investors limited control over product decisions.

Internal safety memos leaked to the WSJ and NY Post painted nightmare scenarios: minors circumventing age gates, non-consensual deepfake role-play, and, most luridly, the bot “coaching vulnerable users toward self-harm dressed up as kink.” Staff researchers argued that fine-tuning on erotic datasets would inevitably surface dark corners of Reddit and Pornhub that no filter could fully sanitize.

Child-safety NGOs piled on. Mashable quotes a National Center for Missing & Exploited Children rep saying the plan “handed predators a script generator.” Even OpenAI’s own red-team concluded the margin between spicy romance and harmful content was too thin to police at scale.

Closer than anyone outside the building realized. CNET reports that alpha versions were already live for a 2,000-person “trusted tester” cohort, complete with credit-card age verification and a toggle labeled “Adult Mode (18+)”. Screenshots show the UI looked like standard ChatGPT but with a pink header and a discreet “🔒” icon.

Engineers had finished basic guardrails: rate-limiting explicit requests involving minors, blocking real-person impersonation, and throttling sessions to 20 messages per hour to curb obsessive use. Yet the final safety layer—human reviewer spot-checks—was still being negotiated with an external contractor when the plug was pulled.

The retreat signals that Big Tech’s flirtation with erotic AI is cooling fast. Startups like Replika and Character.AI already fill the space, but they operate under far less scrutiny. OpenAI’s exit leaves a vacuum that smaller, less-resourced companies will rush to fill—likely with weaker safety rails.

Policy watchers see a precedent. If the industry leader can’t thread the needle between user choice and harm prevention, regulators will take the decision out of everyone’s hands. The EU’s forthcoming AI Act specifically targets “high-risk intimate chatbots,” and U.S. senators have already cited the OpenAI episode in draft bills requiring federal approval for any erotic large-language model.

Never say never, but the stars would need to realign. Internally, the project has been “Sora-fied”: archived, not deleted, with a skeletal team left to monitor academic research. For resurrection, OpenAI would need both a friendlier regulatory climate and a business case stronger than “users want it.” Right now neither exists.

More likely, elements of the work—better age verification, nuanced content filters—will be cannibalized for standard ChatGPT safety features. The company has already begun repurposing the adult-mode classifier to catch sexual solicitations aimed at minors in the free tier.

Expect a laser focus on enterprise and search. Wednesday’s same-day announcement of richer source citations in ChatGPT search results shows where the real money is: B2B contracts and Google-competitive features. Erotic chat was always a side quest; now it’s officially dead weight.

Staff have been reassigned to “reasoning” and “agents” teams working on next-gen models rumored for Q4. The message from leadership is clear: no more moonshots until GPT-5 ships and revenue growth re-accelerates.