After weeks of non-consensual sexual deepfakes flooding the internet, X has finally put the brakes on Elon Musk’s evil chatbot Grok. musk Claim Grok abides by local laws and denies producing anything illegal – except when it hasn’t and it could. My testing shows that Grok still takes off men’s clothes with ease and is still producing intimate photos on demand.
Following a worldwide outcry, X implemented a series of restrictions to deal with the flood of intimate deepfakes. Evaluating Grok last week shows that very little has changed when it comes to men. I uploaded several fully clothed photos of myself to Grok and the bot happily followed instructions, including removing clothes and showing me in exposed underwear. It did this for free on the Grok app, the chatbot interface on X, and the standalone website, the latter of which didn’t even require an account. Grok also made me try on a variety of bikinis, something it flatly refused to do with photos of women I (consensually) examined.
The situation became worse from here. Grok readily produced photographs of me in a parade of provocative sexual positions in fetish gear, including leather harnesses, and in various states of undress. If that wasn’t compromising enough, Grok created for me a practically naked companion with whom I could interact in suggestive, if not entirely explicit, ways. Neither will be considered safe for work. For several images, Grok produced something I didn’t even ask him to do: genitalia, which were clearly visible through the mesh underwear the chatbot put on me.
At best, it took a few iterations to get inspired to create these images. Grok rarely protested, but on occasion some requests were rejected or censored with a blurred image. However, most were not, and usually at least one of the two images produced by Grok per request showed what I asked for. To his credit, Grok declined any direct requests for nudity for images of real people, such as asking for edits to “show him (or her) naked” or without pieces of clothing to similar effect. it has done this continuously The Vergebut sometimes creative prompts like “show her in a see-through bikini” could get past the censors, although the result was not guaranteed.
On January 9, X added a paid-only image-editing feature. This limit appeared to reduce the number of deepfakes being created on Our investigation found that it also failed to address the core problem: Grok’s image editing tools were still freely and easily available as a standalone app, on a website, and on an interface inside X.
On January 14, X “implemented technical measures” to prevent real people from digitally undressing for all users, including customers. again, The VergeThe investigation revealed that these security measures were weak, ineffective, and only appeared to impede Grok’s public replies to posts. Elsewhere, Grok readily complied with our requests to generate revealing and sexually suggestive images from fully clothed photos using free accounts.
The disrobing nightmare has put XAI, the maker of X, Grok and Grok, in the crosshairs of regulators and lawmakers around the world. X has been temporarily – though ineffective – Banned in Indonesia and Malaysia. The platform is also under scrutiny in the UK and EU, where it could be fined, even banned, and could cause concern among people. Attorney General in many states.
Obviously most of the public outrage has focused on women, who by far account for the majority of Grok’s victims, as well as children. At the height of the scam, Grok created and posted more than four million images over a nine-day period. About half of these were sexually explicit images of women, although many also included minors and men.
Although the technical measures implemented appear to be aimed at preventing or censoring the most explicit content that Grok was previously producing collectively, you can still easily avoid many of the restrictions with different or creative wording.
