Openai kills its erotic chatbot to dodge regulatory heat
OpenAI has yanked the plug on its long-rumoured “spicy” chat mode, turning a titillating product tease into a sober lesson on what even the fastest-moving AI lab will no longer touch. The decision, confirmed by the Financial Times, shelves indefinitely any plan to let paying adults swap steamy text with a fine-tuned version of ChatGPT.
Inside the company the project had already earned a nickname—“ChatXXX”—and product decks showed healthy revenue forecasts from lonely hearts willing to pay premium tokens for after-dark role-play. Engineers had solved the obvious technical riddle: how to keep the model from vomiting illegal or non-consensual scenarios. What they could not solve was the reputational Sudoku that follows once regulators, child-safety groups and Fortune-500 partners start asking hard questions in the same week.
The revenue was tempting, the risk profile lethal
Sources close to the board say the final slide that killed the feature displayed two numbers: a $400 million annual projection on the left, and a 42 % spike in policy-risk mentions on the right. The latter came straight from the software that scrapes parliamentary transcripts, FTC filings and EU docket leaks. When the line for “probable enforcement” crossed the line for “expected ARR,” CEO Sam Altmanreportedly closed the deck and said: “We’re not doing porn with our logo on it. Not now, not ever.”
Investors who once salivated over an untapped adult vertical were reminded that OpenAI’s $80 billion valuation hinges on a story of responsible frontier stewardship. One venture partner put it brutally: “There are easier ways to make money than explaining to Congress why your bot sexted a 14-year-old in Alabama.”
The market gap will be filled by less scrupulous players
Start-ups headquartered in Cyprus and the Cayman Islands are already bidding for the engineers laid off from the adult-mode team. Their pitch: no KYC, no geofence, no content logs. The user only needs a crypto wallet and a Telegram handle. While OpenAI retreats behind its usage-policy moat, the vacuum is being stuffed with open-source models fine-tuned on Literotica archives and sold as “uncensored companions.”
That fragmentation matters. Every time a big brand walks away from the red-light district, the traffic migrates to darker corners where age verification is a check box and consent is whatever the prompt says it is. Regulators may cheer OpenAI’s prudence, yet the practical effect is to push the problem downstream to actors who care not one jot about congressional hearings.
The line drawn today will shape tomorrow’s liability law
By publicly declaring that certain prompts are “outside our risk tolerance,” OpenAI has handed plaintiffs’ lawyers a gift-wrapped precedent. The next time a minor is abused by a rogue AI service, the complaint will cite OpenAI’s own standards as evidence that the harm was “foreseeable and preventable.” In tort law that is the fast lane to punitive damages.
Meanwhile, the EU’s forthcoming AI Act classifies intimate chatbots as “high-risk systems,” triggering conformity assessments that can cost up to €15 million. OpenAI’s pre-emptive strike means it avoids that quagmire entirely; smaller rivals cannot. The result is a two-tier market where compliance costs act like a sin tax on anyone still willing to host digital desire.
The company will book zero direct revenue from the cancelled feature, but it bought something more valuable: a pristine brand at the moment Washington and Brussels write the rules. In the high-stakes poker of AI governance, folding early on porn was the cheapest ante OpenAI could pay to stay at the table for the bigger hands ahead.