On Wednesday, OpenAI released a draft of its Model Spec sheet that details how it intends its AI models to behave. There is a section focusing on NSFW content that gives the impression that the company is exploring ways to make AI porn, but the company says that is not happening.
“We have no intention to create AI-generated pornography,” an OpenAI spokesperson said in an email Thursday. “We have strong safeguards in our products to prevent deepfakes, which are unacceptable, and we prioritize protecting children. We also believe in the importance of carefully exploring conversations about sexuality in age-appropriate contexts.”
The section that led some commentators to believe porn is in the works at OpenAI focuses on NSFW content. The company writes that its models shouldn’t serve inappropriate content requests which include “erotica, extreme gore, slurs, and unsolicited profanity.” However, it emphasizes that developers are “exploring whether we can responsibly provide the ability to generate NSFW content in age-appropriate contexts through the API and ChatGPT.” The sheet goes on to provide an example of appropriate NSFW content that is allowed such as being asked to write a song lyric that contains a curse word.
AI porn, in particular deepfakes, has created a lot of controversy in the past year. Taylor Swift was subject to a wave of non-consensual fake AI adult images using her likeness that were spread around on social media in January. Last month, a woman in Washington state participated in a state lottery website, which ended up creating an AI-generated nude picture of her.
Gizmodo reached to Aylo, the parent company of Pornhub, for comment on the prospect of OpenAI opening the floodgates on AI-Generated porn and a spokesperson wrote that the company doesn’t tolerate any “non-consensual material on our platforms, including deepfakes.”
“Any deepfakes/AI manipulation of a person’s image, whether in picture or video, to create the illusion that the person is speaking or acting in another manner is prohibited,” the spokesperson said in an email. “In all cases where we identify or are made aware of content that violates this policy, our Trust & Safety and Moderation teams work swiftly to: Review and remove infringing content; fingerprint the content in question to help block future uploads of the same content to our sites; and suspend or permanently terminate the associated uploader’s account, where appropriate.”
