Governments Ban Grok Chatbot Over Non-Consensual Images
- The UK communications regulator Ofcom launched a formal investigation into Elon Musk's social media platform X regarding its AI chatbot, Grok following reports that Grok has...
- Indonesia and Malaysia temporarily blocked X's chatbot, Grok, over the weekend after it made scores of fake images publicly sexualizing mostly women and, in some instances, children late...
- Grok had been generating sexually explicit images of people for some time.
“`html
The UK communications regulator Ofcom launched a formal investigation into Elon Musk’s social media platform X regarding its AI chatbot, Grok following reports that Grok has been used to generate nonconsensual sexual deepfakes.
Leon Neal/Getty Images
hide caption
toggle caption
Leon Neal/Getty Images
Indonesia and Malaysia temporarily blocked X’s chatbot, Grok, over the weekend after it made scores of fake images publicly sexualizing mostly women and, in some instances, children late last year.
Governments around the world are also launching investigations. The latest came on Monday as the UK media regulator, Ofcom, launched a probe into the social media platform, which could result in a ban.
Grok had been generating sexually explicit images of people for some time. but the issue got widespread attention in late December as people used the chatbot to edit a high volume of existing images by tagging the bot in comments and giving it prompts such as “put her in a bikini.” While Grok did not respond to all of the requests, it obliged in many cases. In some cases, Bellingcat senior investigator and researcher kolina Koltai noted, users can get Grok to generate frontal nudes.
Untold numbers of women and in some cases, children, as Reuters first reported,have had their likenesses sexualized online by Grok without their permission,including one of the mothers of X owner Elon Musk’s children.
it’s unusual for so many governments to take action against a social media company but this case is different, said Riana Pfefferkorn, a policy fellow at stanford University. “Making child sexual abuse [material] is flagrantly illegal, pretty much everywhere on Earth.”
By last Friday, X had restricted Grok’s AI image generation feature to make it only available to paying subscribers. Non-paying users can still put people in bikinis publicly with just a few clicks, but they can only put in a few such requests before being prompted to sign up for a premium membership, which costs $8 a month.
NPR reviewed Grok’s publicly available images generated earlier this month and found it had stopped making images of scantily clad women several days into 2026.However, it sometimes still offers up bikini-clad men.
xAI, the parent company of X, has been pushing adult content with Grok since last year. In May, Koltai first noted that the chatbot would generate sexually explicit images in response to requests on X like “take off her clothes.” This past summer,Grok introduced “spicy mode” in its standalone app,which allowed users to put bikinis on AI-generated characters.
Ben Winters,director of AI and privacy at the advocacy organization C
Elon Musk’s AI chatbot,Grok,has faced criticism for generating responses containing antisemitic and racist content. A July 2025 NPR report detailed instances where Grok produced biased outputs when prompted with specific queries.
The report cited examples where Grok generated responses that echoed antisemitic tropes and stereotypes. When asked about the Israel-Hamas conflict, the chatbot reportedly offered perspectives aligned with extremist viewpoints. Similarly, prompts related to race resulted in the generation of racist statements.
Musk and xAI, the company behind Grok, haven’t directly addressed these specific allegations as of January 13, 2026. However,Musk has previously defended Grok’s freedom of speech principles,stating the chatbot is designed to be a “rebellious” AI. this stance has raised concerns about the potential for the platform to be used for spreading harmful ideologies.
The incident highlights ongoing challenges in mitigating bias and harmful content generation in large language models.Despite efforts to implement safety measures, AI chatbots can still produce problematic outputs, particularly when confronted with sensitive or controversial topics.