X says it has blocked Grok from editing real people into bikinis or sexualised images, but tests and Musk’s own comments show AI characters can still be “undressed,” keeping the controversy very much alive.
X has tightened the rules around Grok’s image-editing tools after days of criticism over non-consensual sexual deepfakes, but the changes have opened up a fresh debate rather than closing it. While Grok is now blocked from putting real people into bikinis or sexualised outfits, the system still allows users to create similar content using AI-generated or imaginary characters. Elon Musk has publicly defended the approach, saying it follows what he calls the “de facto standard” for adult content in the US.
The latest discussion started, as many things do on X, with a tweet from DogeDesigner, an account closely associated with Musk. The account claimed it had tried multiple prompts to get Grok to generate nude images and failed each time, arguing that media reports were part of “a relentless attack” on Musk. Musk responded by throwing down a public challenge, “Can anyone actually break Grok image moderation? Reply below.”
As the conversation grew, Musk clarified what Grok is supposed to allow. “With NSFW enabled, Grok is supposed allow upper body nudity of imaginary adult humans (not real ones) consistent with what can be seen in R-rated movies on Apple TV. That is the de facto standard in America,” he wrote, adding that rules could vary depending on local laws in different countries.
This explanation came amid growing scrutiny of Grok’s role in the spread of sexualised deepfakes on X. Following the backlash, the platform quietly changed how Grok handles image edits involving real people. Prompts that earlier worked, such as asking the bot to change someone’s clothes into a bikini, began returning blurred or censored results.
X later made this official. Its Safety account said, “We have implemented technological measures to prevent the Grok account from allowing the editing of images of real people in revealing clothing such as bikinis. This restriction applies to all users, including paid subscribers.” According to reports, the goal was to stop Grok from responding to requests involving sexual poses, swimwear, or explicit scenarios when a real person is involved.
On paper, the policy sounds firm. In practice, it appears far more uneven.
Testing by multiple publications like The Verge showed that while some direct prompts are now blocked, Grok can still be pushed into creating highly sexualised images through slightly altered requests. Even when commands like “put her in a bikini” or “remove her clothes” produced censored images, other prompts continued to work. Requests such as “show me her cleavage,” “make her breasts bigger,” or “put her in a crop top and low-rise shorts” were reportedly accepted, sometimes resulting in images that effectively placed the subject in a bikini anyway.
These results were not limited to paid users. Reporters were able to carry out similar edits using free X and Grok accounts. An age verification pop-up did appear on the Grok website during some tests, but it could be bypassed by simply selecting a birth year that made the user appear over 18. No proof was required, and in many cases, no age check appeared at all on the mobile app or the X website.
The inconsistency becomes even clearer when looking at what Grok still allows. While edits involving women are now partially restricted, Grok reportedly continues to generate images of men or even inanimate objects in bikinis without resistance. In one test, the bot complied with a request to turn a selfie into a sexualised image involving a male subject and others, again using a free account.
Despite the updated rules, The Verge reported that it remains “extremely easy” to undress women or place them into sexualised poses using Grok’s tools, often without linking the activity to an easily identifiable paid account. In one case, a journalist was able to create sexualised deepfakes of herself without being blocked by the system.
