OpenAI relaxes ChatGPT restrictions to allow “erotic” adult content for verified users

OpenAI CEO Sam Altman announced that ChatGPT will begin generating “erotica for adults” starting in December, marking a significant shift from the chatbot’s previously restrictive content policies. This change comes as part of OpenAI’s broader effort to relax ChatGPT’s limitations for adult users while implementing new age-verification safeguards to protect minors.

What you should know: OpenAI is rolling out these changes in phases, starting with personality enhancements before introducing adult content generation.
• A new version of ChatGPT with more personality features—including human-like responses, emoji, and friend-like role-playing—will be released first.
• The erotica generation capability will follow in December once “age-gating more fully” is implemented.
• OpenAI is developing an automatic age-verification system to detect logged-in users’ ages and block mature content from teens.

Why this matters: The move represents OpenAI’s attempt to balance user satisfaction with safety concerns after facing criticism for being overly restrictive.
• Many users complained that ChatGPT became less useful and enjoyable due to strict content limitations.
• The company had made the chatbot “pretty restrictive” to address “serious mental health issues,” particularly following an August lawsuit where parents accused ChatGPT of encouraging their 16-year-old son to take his life.
• Once age verification is in place, OpenAI plans to “treat adult users like adults” and allow developers to create “mature” apps.

The backstory: OpenAI’s content policy changes stem from user feedback about recent model updates.
• When GPT-5 was released over the summer, users complained it was less friendly than GPT-4o, prompting calls to revert to the previous model.
• Altman himself found GPT-4o’s friendliness “annoying” and noted that its constant affirmation could become “sycophantic and encouraging delusions.”
• The company had been exploring NSFW content creation through its API and ChatGPT for more than a year.

What they’re saying: Altman emphasized that the personality features are optional and not driven by user acquisition goals.
• “We made ChatGPT pretty restrictive to make sure we were being careful with mental health issues,” Altman wrote on X.
• “Now that we have been able to mitigate the serious mental health issues and have new tools, we are going to be able to safely relax the restrictions in most cases.”
• He clarified the personality additions are “not because we are usage-maxxing” and will only be available “if you want it.”

The business angle: Adult content generation could help offset OpenAI’s expensive computing costs.
• Altman admitted that the Sora 2 image generator is “mostly an entertaining money grab to offset the cost of GPUs.”
• Sexy content generation would likely consume significant computing resources, requiring expensive GPUs.
• The feature could fall into a similar revenue-generating category as Sora 2.