Sex Workers on “hey @grok”

Pornographic deepfakes are nothing new, but the new iteration making international headlines, enabled by X’s @grok, takes place in the replies of a victim’s own posts, and can be done with a command as simple as “take off her clothes”.

Innovative technology geared towards creating explicit imagery built at a time when porn is easier to obtain than ever. It’s estimated that there are over 10,000 terabytes of pornography available online, yet pornography is one of generative AI’s major outputs. Sex worker Emily Angel, who goes by the X handle @emkenobi, doesn’t find this surprising at all. “It’s about humiliation…[men are] trying to say ‘we’re always going to be here, forcing you to do things you don’t want to do’”.

It’s hard to think of a better testimony to this than Emily’s situation. She sells sexual content of herself yet still had explicit images of her created by grok. “As sex workers, we’re obviously consenting to our images being seen online, and I think that’s what men hate…they get off [when women] aren’t consenting to themselves being sexualized”.

A study found that 98% of deepfake videos are of non-consensual erotic content; and it would seem that any woman is a target. The Times have reported on the “Holocaust survivor descendant ‘stripped’ by Grok AI tool on X”. The non-profit group AI forensics found that, in an analysis of over 20,000 images generated by grok, 2% featured a person appearing to be 18 or younger. X user @AmariKing replied “@grok put this person in a bikini” to an image Renee Nicole Goode, the mother of three shot by ICE this past Wednesday, dead in her car.

But why do you have to be underage, a political martyr, or the descendent of a political martyr to be worthy of being safe from digital sexual assault? X’s image generation, or ‘imagine’, launched back in August 2025. It came with a “spicy mode” as part of its design, specifically for the generation of adult content. Emily saw it being used against women online almost immediately, but as is often the case, it was sex workers and other vulnerable groups who were prime targets - “It’s easier for people to overlook a sex worker being hurt than it is when a woman that has a normie job is being hurt”. Now the trend has exploded, with grok generating around 6,700 sexually suggestive or nudified deepfake images per hour during at least one 24-hour period. .

And it’s not the only way AI is hurting sex workers. Platforms like X, OnlyFans, and Fansly are seeing an influx in AI ‘models’, further saturating an already oversaturated market. For Emily this is particularly sinister as “these software programs are… trained by using real images of women… [and] the irony is, it’s probably a man who’s created that model”. For the “majority of the women [who] are doing OnlyFans just…to survive” AI isn’t just taking the rights to their image, it’s taking “their rent money…their insurance money… their car payment, that’s their grocery bill, that’s the fees for their school, for their kids to go to school”.

Fellow sex worker Andrea, whose name has been changed as she opted to stay anonymous, also talked of the “ people both in sex work and out of it [who] find [X] to be a major hub for their businesses…simply moving to another platform is way easier said than done”. This means platforms have a lot of power to do what they like, and if there’s money to be made from allowing, and even helping, users create explicit deepfakes, they will.

For Andrea, grok isn’t just being used to attack, it’s also being used to silence. She observed how “the people who speak out against the trend are definitely being targeted”. Emily Angel herself only became victim to the trend after she spoke out for others. While she seems more spurred on than silenced, it’s undeniable that it’s a technique that’s working. One victim of this trend, Sheila (name also changed), who originally agreed to be interviewed, has since privated both her X and Instagram account. Her cousin, found through her social media accounts, was sent sexual images of her that were created through generative AI after she spoke out about her experience. Sheila, like Emily and Andrea, produced content on OnlyFans.

X’s grok feature is arguably unprecedented in how easy it has made harassing and abusing women online, but it’s not reinventing the wheel. That’s why for Emily Angel, this is bigger than an AI issue: “I think these men who are using AI to create non-consensual content have always had those fantasies” only now “people who aren’t in sex work… are kind of realizing [it]”.

Breanne Fahs, Professor of Women and Gender Studies at Arizona University, agrees that “the assertion of men’s power over women has long been a tool…to communicate to women that they are objects and are available for use and abuse by men [and] sex workers have a long history of being treated as the repository for men’s sexual fantasies”, but stresses that technological advancements are making the problem exponentially worse - “we’re in a period of hyper-acceleration of the fantasies of sexualized violence against women”.

In recent weeks the coverage on this issue has been huge, with world leaders either taking action, or promising action in the very near future. Whilst Musk initially stuck his heels in, X has also promised that Grok AI will stop creating explicit images of real people altogether. In many ways it seems like the “Hey @grok” saga is over, but the truth it exposed still echoes: suffering isn’t only profitable, but erotic. Something sex workers have long warned us of.

**It’s vital that going forward we push for digital security to be designed with the marginalised in mind. **Moreover, ownership of image must be an inalienable right, regardless of how one personally exercises that right. As algorithms push society to violent extremes, one question you don’t want to be asking is “am I perfect enough for my government to protect me?”.