OpenAI CEO Sam Altman announced in a publish on The company will soon loosen some of ChatGPT’s security restrictions on Tuesday, allowing users to make the chatbot’s responses friendlier or more “human” and for “verified adults” to engage in erotic conversations.

“We made ChatGPT quite restrictive to ensure we were alert to mental health issues. We realize this made it less useful/enjoyable for many users who didn’t have mental health issues, but given the severity of the issue we wanted to fix this,” Altman said. “In December, as we implement age restrictions more fully and as part of our principle of “treating adult users like adults,” we will allow even more, such as erotica for verified adults.”

The announcement represents a major pivot in OpenAI’s months-long work to address the troubling relationships some mentally unstable users have developed with ChatGPT. Altman appears to declare an early victory over these issues, claiming that OpenAI has “been able to mitigate the serious mental health issues” linked to ChatGPT. However, the company has provided little to no evidence of this and is now moving forward with plans for ChatGPT to engage in sexual chats with users.

This summer, several troubling stories have emerged around ChatGPT, particularly its GPT-4o model, suggesting that the AI ​​chatbot could be leading vulnerable users down delusional rabbit holes. In one case, ChatGPT appeared to do this convince a man that he was a mathematical genius that he needed to save the world. In another, the parents of a teenager sued OpenAIalleging that ChatGPT encouraged her son’s suicidal ideation in the weeks leading up to his death.

In response, OpenAI has released a series of security features to address AI sycophancy: the tendency of an AI chatbot to hook users into accepting whatever they say, even negative behaviors.

Launch of OpenAI GPT-5 in August, a new AI model that shows lower rates of sycophancy and features a router that can identify user behavior. A month later, OpenAI launched child safety features, including an age prediction system and a way to parents to monitor their children’s ChatGPT account. OpenAI announced Tuesday expert advice of mental health experts to advise the company on wellbeing and artificial intelligence.

Just a few months after these stories emerged, OpenAI appears to believe that ChatGPT’s issues related to vulnerable users are under control. It’s unclear whether users are still falling down delusional rabbit holes with GPT-5. And although GPT-4o is no longer the default in ChatGPT, the AI ​​model is still available today and used by thousands of people.

OpenAI did not respond to TechCrunch’s request for comment.

The introduction of erotica into ChatGPT is uncharted territory for OpenAI and raises broader concerns about how vulnerable users will interact with the new features. While Altman insists that OpenAI isn’t “maximizing usage” or optimizing engagement, making ChatGPT more erotic could certainly attract users.

Allowing chatbots to engage in romantic or erotic role-play has been an effective engagement strategy for other AI chatbot providers, such as Character.AI. The company has gained tens of millions of users, many of whom use its chatbots at a high rate. Character.AI said in 2023 that users spent on average two hours a day talking to its chatbots. The company is too deal with a lawsuit on how it handles vulnerable users.

OpenAI is under pressure to grow its user base. While ChatGPT is already used by 800 million weekly active users, OpenAI is competing against Google and Meta to create mass-adopted AI-powered consumer products. The company has also raised billions of dollars to build historic infrastructure, an investment that OpenAI will eventually have to repay.

While adults are definitely have romantic relationships with AI chatbotsit is quite popular even for minors. A new report from the Center for Democracy and Technology found this 19% of high school students you’ve had a romantic relationship with an AI chatbot or know a friend who has.

Altman says OpenAI will soon enable erotica for “verified adults.” It’s unclear whether the company will rely on its age prediction system, or some other approach, to age-restrict ChatGPT’s erotic features. It’s also unclear whether OpenAI will extend erotica to its AI-powered voice, image and video generation tools.

Altman says OpenAI is also making ChatGPT friendly and erotic thanks to the company’s “treat adults like adults” principle. Over the past year, OpenAI has moved towards a more lenient content moderation strategy for ChatGPT, allowing the chatbot to be more forgiving and offer less rejection. In February, OpenAI committed to representing more political viewpoints in ChatGPT, and in March the company updated ChatGPT to allow AI-generated images of hate symbols.

These policies appear to be an attempt to make ChatGPT’s response more popular among a wide variety of users. However, vulnerable ChatGPT users can take advantage of safeguards that limit what a chatbot can interact with. As OpenAI races toward one billion weekly active users, the tension between growth and protecting vulnerable users will only increase.