Adult content creators face deepfake exploitation and AI training risks
A report from MIT Technology Review highlights how deepfakes and AI training data are creating new forms of sexual violence and financial precarity for adult industry workers.
Adult content creators are increasingly reporting that their bodies are being used without consent in nonconsensual intimate imagery, often with celebrity faces superimposed onto their performances. According to a recent report from MIT Technology Review, this practice is described by experts as a new form of sexual violence causing significant psychological distress. The issue extends beyond the individuals whose faces are featured, raising urgent questions about the rights and safety of the performers whose bodies are utilised in these digital reproductions.
Experts warn that generative AI models are likely trained on vast amounts of online adult content, threatening creators' livelihoods and rights. This training data allows AI to replicate how bodies look, move, and perform, potentially replacing human workers. While legal tools such as the Digital Millennium Copyright Act are used for takedowns, enforcement is difficult due to anonymous hosting and jurisdictional issues. New legislation, such as the US Take It Down Act, risks inadvertently removing legal adult content or being weaponised to report lawful material as nonconsensual.
The psychological impact on performers is severe, with experts terming the damage "embodied harms." One creator, identified as Jennifer, described the experience of seeing her body used in a deepfake as traumatic, noting that it feels akin to being part of someone else's abuse. The uncertainty of not knowing what one's body is doing online can be highly unsettling, leading to body dysmorphia and self-censorship. Unlike the face-swapped entities that occupy the uncanny valley, the bodies used in these deepfakes are often real people whose consent was never obtained for such digital exploitation.
To combat this, some creators are exploring contracts for AI duplicates, though enforcement remains challenging. For instance, some performers have signed with platforms hosting AI-generated duplicates to assert ownership, but platform stability is a concern, as seen with the shutdown of Spicey AI. Additionally, companies like Takedown Piracy have added digital fingerprinting technology to help identify and remove problematic videos, even when faces are swapped. However, US copyright violations can be challenging to prove if a body lacks distinguishing features, leaving many creators without clear legal recourse.
The situation is further complicated by the rise of nudify apps and AI-generated bodies, which are becoming more prevalent. These tools allow users to upload clothed photos to generate fake naked images, often drawing on terabytes of online porn. Experts note that while it is difficult to prove specific content is used for training, the ubiquity of adult content makes it a reasonable assumption that it is being used. This has led to calls for legitimate, effective laws that protect content creators from having their likeness abused to train and create AI, as current measures often fail to address the root causes of this exploitation.
