Adult Mode: ChatGPT Can Soon Also Do Erotica

First came the youth protection and the tightening of guardrails for conversations with ChatGPT. Now OpenAI is backtracking in some respects and even allowing erotic content. However, you have to be an adult for that. Overall, the chatbot will apparently react more nuanced to the respective interlocutor in the future.

Sam Altman, CEO of OpenAI, writes about the changes on X: "In December, as we roll out age-gating more fully and as part of our "treat adult users like adults" principle, we will allow even more, like erotica for verified adults." How far "erotica" goes, whether just a little flirting or even sexting will be possible, is unclear.

It is also unclear how one can verify oneself. The observance and control of age limits have always led to discussions on the internet. As a rule, classification is based on self-declaration by users. Social media platforms like Instagram also monitor consumed content and online behavior and classify accounts as teen accounts if suspected.

ChatGPT is to become more personal

OpenAI has apparently also heard the complaints of many people who want to have relationships with the chatbot. After the introduction of GPT-5, there were complaints that the chatbot was reacting much less humanly and was less approachable. OpenAI then already brought the previous model GPT-4o back into the chat. Now Altman has announced a new version that again corresponds more to a personality and the behavior "that people liked about 4o". ChatGPT should react like a friend if desired and, for example, answer very humanly or use lots of emojis. In contrast to social media, Altman emphasizes that the goal is not to keep people on the platform. OpenAI does not yet earn money from advertising shown to users.

The background for the stricter guardrails and the introduction of youth protection were reports about people who have unhealthily close relationships with a chatbot. The case of a 16-year-old, who is said to have committed suicide, is particularly dramatic. The parents are suing OpenAI for a kind of complicity by the chatbot. But according to Altman, the stricter guidelines have led to "some people without mental health problems" finding ChatGPT less helpful and fun. Be that as it may, OpenAI is now able to recognize serious psychological problems and, thanks to new tools, relax the restrictions in most cases. Here too, Altman is not getting more specific about how exactly the chatbot's reactions are controlled.

OpenAI is not alone in the field test of providing adults with erotic chatbots. Elon Musk's Grok has explicitly flirty personalities in store. Character.ai is a provider where you can create AI friends.

Since such relationships carry dangers, California already has an AI law that prescribes transparency and safety precautions. In addition, connections to criminal offenses must be reported. Another bill deals with AI chatbots that explicitly have a friendship mode. Other US states are also working on similar laws.

(emw)