OpenAI has updated its Model Spec guidelines, allowing ChatGPT to generate more mature content under specific circumstances. The new policy shift includes erotica and gore in “appropriate contexts,” such as scientific, historical, news, or creative writing.
The changes come after OpenAI explored the possibility of generating NSFW content through its API and ChatGPT in May 2024. The guidelines now state that sensitive content can be generated, but only in contexts where it is suitable.
However, users have reported finding ways to bypass the content filters, allowing them to generate explicit sexual or violent scenarios without warnings. OpenAI notes that its Usage Policies still apply, prohibiting the creation of AI tools for minors with sexual content.
The relaxation of content filters has been met with calls for an uncensored version of ChatGPT, citing limitations in its paternalistic approach. The new policy aims to balance user discretion and responsible content generation.
Source: https://arstechnica.com/ai/2025/02/chatgpt-can-now-write-erotica-as-openai-eases-up-on-ai-paternalism