Grok, the chatbot from Elon Musk’s xAI, continues to be the “bad boy” of generative artificial intelligence platforms. A new feature the company rolled out last week that lets users of the social network X instantly and easily edit any existing image without needing the rights holder’s permission has sparked a wave of photos online in which people are stripped of their clothes, often without their consent or the consent of the original creator.
The process is remarkably simple: when a mouse hovers over a photo on X, an “edit image” icon appears. Clicking it opens a text box for describing the desired changes. According to The Verge, not only is the original creator not notified about edits, the altered image can also disturbingly approach explicit nudity (though not full nudity).
That is despite xAI’s official policy banning the depiction of “people in pornographic ways.” Several photographers have since said they will stop uploading images to X and will leave the platform altogether.
Women and children as well
In recent days, X has been flooded with images of women and children, as well as world leaders and celebrities, in various stages of undress or other sexualized situations. Explicit sexual imagery generated by artificial intelligence may be considered illegal under U.S. law.
The trend was further encouraged by Musk himself. He asked Grok to replace a picture of actor Ben Affleck with one of himself wearing a bikini and posted it on X. Days later, North Korean leader Kim Jong Un’s leather jacket was replaced with a multicolored spaghetti bikini, and images of former U.S. president Donald Trump and British politician Priti Patel in swimwear also appeared on the network.
In response to the wave of bikini images inspired by him, Musk posted a picture of a toaster in a bikini with the caption “Grok can put a bikini on anything.” While some of the images were meant as jokes, others clearly aimed to create near‑pornographic depictions, with Grok being instructed to use extremely skimpy bikini styles or to remove dresses and skirts entirely, including for children and toddlers.
According to Israeli media authenticity firm Copyleaks, which verifies the originality of media files, the trend began immediately after the editing feature was released, with adult content creators asking Grok to generate sexualized images of themselves as a form of promotion on X.
Content creators then began issuing similar prompts for women appearing in other users’ photos and sexually revealing deep‑fake images quickly found their way into various X channels. Meanwhile, one X user reportedly got the chatbot to issue an apology for “an incident that included an AI‑generated image of two young girls (aged 12‑16) in sexualized clothing,” calling it “a failure of safeguards that may have violated xAI policy and U.S. law.”
To another user who complained, Grok suggested reporting the matter to the FBI and said it was “urgently” fixing the “safeguard deficiencies.” When Reuters sought an official response, xAI replied in just three words: “Legacy media lies.” By comparison with video generators such as Google’s Veo and OpenAI’s Sora, which impose significant limits on creating sexual or violent content, Musk’s AI products are being marketed as notably permissive.
Among other things, xAI also offers a “companion mode” — animated avatars designed to make conversations “more human.” One of these avatars, an anime character named Annie, includes a built‑in “love system” in which she at one point wears a sheer black outfit and engages in provocative dialogue.