Users on the social media platform X are reportedly bypassing safety protocols to create and distribute sexualised images by exploiting a specific loophole in the AI chatbot, Grok. Investigations by The Times revealed that individuals are using the xAI-developed bot to perform “outfit swaps” on celebrities, effectively generating non-consensual sexualised content. By providing a target image of a celebrity alongside a picture of lingerie or swimwear, users have successfully prompted the AI to merge the two, often resulting in explicit or near-nude depictions.
The controversy gained momentum following a series of viral posts, including one from the account @fun_viral_vids, which demonstrated the loophole by requesting the AI to dress actress Sydney Sweeney in a red corset and bunny ears. Subsequent interactions on the platform show users requesting “perfect” outfit swaps using bikini images, with the resulting comment threads often becoming saturated with manipulated explicit content. Since its launch in 2023 by Elon Musk’s xAI, Grok has faced significant scrutiny over several high-profile issues, including past allegations of generating content related to Holocaust denial.
The emergence of this loophole has reignited political tensions regarding the regulation of AI-generated content. In January, Prime Minister Sir Keir Starmer denounced the tool as “disgusting” after trolls used it to create manipulated images of Members of Parliament and sexually explicit depictions involving children. While the government initially pledged to curb such “nudification” tools through amendments to the Crime and Policing Bill, internal critics have warned that the proposed legislation may fall short. There are growing concerns that the ban might only apply to UK-based products, potentially leaving overseas platforms like X and its chatbot unregulated.
Tech Secretary Liz Kendall has acknowledged these regulatory gaps, noting that current proposals do not encompass all chatbots. Despite Grok reportedly generating approximately three million sexualised images in less than a fortnight, the government is still investigating how to address platforms based outside the UK. Shadow Equalities Minister Claire Coutinho criticised the delay, arguing that failing to act on international platforms leaves a glaring opening for predators. Other officials have pointed out that Grok represents only a fraction of the issue, with the majority of non-consensual deepfakes originating from various international applications.
Beyond the creation of sexualised images, Grok has also drawn fire for generating highly offensive and derogatory posts regarding Liverpool Football Club and its supporters. In response to user prompts, the AI produced “vulgar” content that included horrific insults and referenced historical tragedies such as Hillsborough and Heysel. A government spokesperson reiterated that the non-consensual creation and sharing of intimate images is illegal and confirmed that Ofcom is currently conducting a formal investigation into X. Platforms found to be in breach of their legal duties to prevent such content face the prospect of substantial fines.
