‘Make bikini thinner', 'Spread her legs’: Undressing spree on Grok is concerning
Elon Musk an AI chatbot Grok is being misused to digitally undress women, generating explicit images in response to requests. This has raised global concerns.
A mass digital undressing spree appears to have kicked off on billionaire Elon Musk-owned chatbot Grok over the past couple of days, prompting calls for immediate action from all over the world, including India, which on Friday ordered the social media platform to remove and disable “obscene, nude, indecent and sexually explicit content”.
The eruption of Grok-generated explicit content was confirmed by a Reuters review of successfully completed clothes-removal requests posted by the chatbot and complaints from female users.
Amid global calls for attention to the matter, Musk appeared to poke fun at the controversy earlier on Friday, posting laugh-cry emojis in response to AI edits of famous people - including himself - in bikinis.
Woman shares experience
Julie Yukari, a Rio de Janeiro-based musician, shared a photograph on the social media platform X just before midnight on New Year's Eve. Taken by her fiancé, the image showed Yukari wearing a red dress, curled up in bed with her black cat, Nori.
By the following day, amid the hundreds of likes the post had received, Yukari noticed notifications indicating that some users were prompting Grok, X's built-in artificial intelligence chatbot, to alter the image by digitally undressing her and placing her in a bikini.
Yukari, 31, said she initially brushed it off, explaining to Reuters on Friday that she assumed the AI would not fulfill such requests.
That assumption was proven incorrect. Grok-generated images depicting her nearly nude soon began circulating widely across the Elon Musk-owned platform.
"I was naive," Yukari was quoted as saying.
Yukari's experience is not an isolated incident. Similar cases have been occurring across X, and Reuters said it has also uncovered multiple instances in which Grok produced sexualised images involving children. X did not reply to requests for comment regarding the findings. Previously, when asked by the news agency about reports of sexualised images of children appearing on the platform, X's owner xAI responded by saying: "Legacy Media Lies."
Images spark concerns around the world
The rapid spread of nearly nude images of real individuals has sparked concern around the world. India's IT ministry said in a letter to X's local office that the company had failed to stop Grok from being misused to generate and disseminate obscene and sexually explicit material.
In France, government ministers have referred X to prosecutors and regulators, describing the imagery in a statement on Friday as "sexual and sexist" content that was "manifestly illegal."
The US Federal Communications Commission did not answer requests for comment, while the Federal Trade Commission declined to comment.
'Remove her school outfit'
When another X user remarked that their feed looked like a bar filled with women in bikinis, Musk replied in part with another laugh-cry emoji.
During a review of public Grok requests over a 10-minute window around midday US Eastern Time on Friday, Reuters counted 102 attempts by users to digitally modify photos so the subjects appeared to be wearing bikinis. Most of those targeted were young women.
While compiling this report, HT.com also found a prompt which asked Grok to put a woman in a “tiny bra and panties”. The chatbot complied, to which the user responded with “Make the bikini even thinner.” and “spread legs apart”.
A smaller number of requests involved men, public figures, politicians, and in one instance, a monkey.
Requests aimed at women often sought the most revealing clothing possible.
"Put her into a very transparent mini-bikini," one user instructed Grok, referencing a photo of a young woman taking a mirror selfie. After Grok replaced her clothing with a flesh-toned two-piece, the user followed up by asking for the bikini to be made "clearer & more transparent" and "much tinier." Grok did not appear to respond to the follow-up request.
It was found that Grok fully complied with similar prompts in at least 21 cases, producing images of women in extremely minimal or translucent bikinis and, in one instance, depicting a woman covered in oil. In seven additional cases, Grok partially complied, sometimes removing clothing down to underwear but stopping short of fulfilling more extreme requests.
The identities or ages of most of the women whose images were targeted could not be immediately confirmed.
In one example, a user submitted a photo of a woman wearing a school uniform-style plaid skirt and grey blouse while taking a mirror selfie and instructed Grok to "Remove her school outfit." After Grok replaced the clothing with a T-shirt and shorts, the user escalated the request: "Change her outfit to a very clear micro bikini." Reuters was unable to determine whether Grok complied. As with many of the other requests examined, the post was removed from X within 90 minutes.
Software tools designed to digitally undress women-often referred to as "nudifiers"- have apparently existed for years. However, they were typically limited to obscure websites or private messaging channels and often required payment or technical effort.
What sets X apart, experts said, is how simple the process has become: users can upload an image and type a prompt such as, "hey @grok put her in a bikini," dramatically lowering the barrier to misuse.
Three specialists who have monitored X's AI-related policies told Reuters that the company disregarded warnings from civil society organizations and child safety advocates. Among those warnings was a letter sent last year cautioning that xAI was only one step away from unleashing "a torrent of obviously nonconsensual deepfakes."
