Elon Musk’s AI chatbot, Grok, will no longer be able to modify photos of real individuals to depict them in revealing clothing in regions where such activity is illegal. The decision follows widespread criticism over the use of AI to create sexualised deepfake images.
In a statement posted on X, the company confirmed it had introduced technical safeguards to prevent Grok from editing images of real people into explicit or suggestive content.
The UK government described the move as a “vindication” of its calls for stronger oversight of the platform, while communications regulator Ofcom welcomed the change but confirmed its investigation into possible legal breaches is still ongoing.
Despite the update, campaigners and victims argue that the damage has already been done. Ofcom said it is working urgently to understand how the issue occurred and what steps are being taken to prevent future abuse.
Technology Secretary Liz Kendall welcomed the decision but stressed that Ofcom’s investigation must fully establish the facts.
Journalist and activist Jess Davies, whose image was altered using Grok, said the changes were a step in the right direction but criticised X for allowing such content in the first place. She described the platform’s overall response as inadequate and expressed concern about the number of women affected by AI-generated abuse.
Dr Daisy Dixon, a philosophy lecturer at Cardiff University, previously told the BBC that having her image manipulated left her feeling humiliated and fearful. She called the platform’s reversal a partial victory but emphasised that the harm to victims remains significant.
Andrea Simon, director of the End Violence Against Women Coalition, said the move showed that pressure from victims, campaigners, and governments can force tech companies to act. However, she urged platforms to take proactive steps to prevent future abuse as AI technology continues to evolve.
The announcement came shortly after California’s Attorney General confirmed an investigation into the spread of sexualised AI-generated images, including those involving children.
X stated it is now blocking users in certain regions from creating images of real people in bikinis, underwear, or similar attire through Grok. The company also confirmed that only paid users can access image-editing features, which it says helps ensure accountability.
With NSFW settings enabled, Grok may still generate limited nudity involving fictional adult characters, in line with US film standards, Musk explained. However, he noted that content rules vary depending on local laws.
Earlier, Musk had defended the platform, claiming critics were attempting to restrict free speech, and shared AI-generated images of UK Prime Minister Sir Keir Starmer in a bikini.
It remains unclear how effectively X will enforce regional restrictions or whether users could bypass them using tools such as VPNs, which allow people to appear as if they are accessing the internet from another country.
Similar spikes in VPN usage were seen in the UK last year after new age-verification rules were introduced for adult websites under the Online Safety Act.