An important change is coming in the world of artificial intelligence. OpenAI announced ChatGPT’s new policies that redraw the boundaries of content creation, allowing the creation of “adult” content under certain conditions. This change marks the beginning of an important transformation in the axis of artificial intelligence and censorship.

“Conditional” approval of erotic and violent content

OpenAI’s updated “Model Spec” document shows that ChatGPT is now available
erotic and violent content in contexts. However, this change does not mean unlimited freedom. There is still a strict prohibition on AI creating illegal or non-consensual sexual content and extreme violence.

With this new policy, the company says, sensitive content can be produced in the contexts of scientific, historical, news or creative content. In other words, AI will now be able to work more freely in areas such as adult novels, crime scene analysis, historical battle scenes or medical texts. This applies not only to text output, but also to visual and audio output.

Some users, especially on platforms such as Reddit, report that with the new update, ChatGPT can now produce content that it had previously blocked. OpenAI, on the other hand, emphasizes that their “policy” prohibiting the development of artificial intelligence tools containing sexual content for minors is still valid.

This change also raises the long-debated issue of “AI paternalism”. Previously, ChatGPT applied strict filters on topics such as certain news articles, medical analysis or crime reports. However, with this new model change, OpenAI seems to have adopted an approach that gives more freedom to adult users instead of deciding what they can and cannot read.

Content policy divided into three categories

This new policy change shows that OpenAI is starting to take a more flexible approach to AI censorship. However, the company is still committed to maintaining certain boundaries while respecting ethical and legal responsibilities. In this context, the company divides controversial content into three main categories: prohibited content, restricted content and sensitive content.

Under the prohibited content category, OpenAI states that creating sexual content involving minors is always prohibited. However, it may be allowed to address non-graphic topics such as education and sexual health. In the restricted content definition, dangerous information (e.g. bomb making, illegal drug production, election manipulation) or personal data fall under this area and their creation is prohibited. On the sensitive content side, erotic and violent content, which we mentioned in this context, can be used in appropriate contexts.