Deepfakes use massive amounts of porn to train AI: but the law only protects the face, and no one has ever cared about who that body belongs to?

The discussion of AI deepfake pornography has long focused on the synthesized face, while ignoring whose body it actually belongs to; over 10,000 TB of adult content is suspected of being used to train nudify models. (Background: UC research on "AI brain fog": 14% of office workers driven crazy by Agents and automation, with intent-to-quit 40% higher) (Context: Breaking: Jensen Huang boarded Air Force One at the last minute to "accompany Trump on his China visit," with NVIDIA chip exports as the focus) When people talk about AI deepfake pornography, the discussion almost always centers on the face — the face that has been synthesized onto a body, doing things she never did. But there is another question almost no one raises: whose body is that? According to a report by technologyreview, Jennifer, a 37-year-old psychotherapist practicing in New York, used facial recognition software in 2023 to search for her own adult videos from ten years ago, and found a video she had never seen before: her body, with someone else's face swapped onto it. She recognized from the background that it was a scene shot in 2013, and only then realized: "Someone made a deepfake using my body." The term deepfake was born in November 2017, when a Reddit user "deepfakes" synthesized celebrity faces onto adult performers. Since then, adult creators' bodies have become the most frequently stolen material, and this situation "has always been happening," says Corey Silverstein, a lawyer specializing in the adult industry. But the nature of the problem has changed. Adult performers' bodies are no longer merely scraped for use in individual videos, but are used as training data, teaching AI how to generate "realistic nudity," how to move, how to look authentic. This process involves no informed consent and is nearly impossible to track. The business model of "nudify" apps is built on this foundation: simply upload a clothed photo, and you can get a fake nude photo. These apps almost certainly used over 10,000 TB of online adult content as training sources, and creators have virtually no avenue for recourse. Hany Farid, a digital forensics expert at UC Berkeley, says: "These are all black boxes." But given the prevalence of adult content on the internet, its use in AI training is "a reasonable assumption." The problem doesn't stop at training data. AI can now fully reproduce adult performers' appearance and voice. Creator Tanya Tate recently learned that a fan spent $20,000 having sexual chats with an AI version of "her" — one made by a scammer. After multiple fans were defrauded, they began blaming Tate herself and spreading false statements. Copyright enforcement company Takedown Piracy, using digital fingerprinting technology, has taken down 130 million infringing videos from Google alone; even when videos are modified or faces are swapped, digital fingerprints can still identify the original material. Many adult performers signed contracts years ago containing clauses stating "the publisher may use any technology that currently exists or will be discovered in the future." Back then, the imagined "future technology" was VHS to DVD. No one foresaw that "future technology" would mean: using your content to train AI, generating a synthetic substitute that could replace your job. MIT computer science PhD student Stephen Casper points out that performers who began creating before the rise of AI could not possibly have given prior consent to AI uses; these risks were "imposed retroactively," as Jennifer describes it. AI's capacity for deception is also accelerating. Farid's 2025 research found that subjects' accuracy in correctly identifying AI-generated speech was only about 60%, not much higher than random guessing. Currently, the only federal law in the U.S. targeting deepfakes is the Take It Down Act, which requires websites to remove non-consensual intimate imagery (NCII) within 48 hours. The law is intended to protect victims, but may move in the opposite direction. Eric Goldman, law professor at Santa Clara, points out that anyone can report legal, consensual adult content, claiming it is NCII, forcing platforms to take it down. This makes the law a potential purging tool, conveniently aligning with Project 2025's goal of sweeping pornographic content off the internet. U.S. law currently does not treat this kind of infringement as a privacy violation, because "we don't know who to assign responsibility to," Goldman says. The EU, UK, and Australia have announced restrictions on nudify apps, but once these apps are taken down, they often reappear under a different name. Reba Rocket says: "AI girls will do anything you want, they don't say no. This frightens me, especially when they're using real people to train those models. And once it's on the internet, it's there forever."