OpenAI has announced a significant update to its guidelines for ChatGPT, allowing the generation of erotica and gore in certain contexts. This policy shift, revealed in the latest “Model Spec” document published on Wednesday, marks a departure from the company’s previous paternalistic approach to content moderation.
Since May 2024, OpenAI has been exploring how to responsibly enable the generation of NSFW content, and the new guidelines specify that while erotica and gore can be created, it must be within appropriate contexts. The document emphasizes that such content should not include illegal or non-consensual activities and should only be generated for scientific, historical, or creative purposes.
Users on platforms like Reddit have already begun testing the relaxed content filters, successfully producing explicit scenarios without warnings. However, OpenAI maintains that its Usage Policies still prohibit the creation of AI tools that include sexual content for minors.
This update aims to alleviate previous restrictions that hindered professionals in fields like law enforcement and legal analysis from utilizing ChatGPT effectively. As OpenAI acknowledges, the need for a “grown-up mode” has never been clearer!
