Grok is now restricted from creating sexualized images of individuals, X Says

On Wednesday, Elon Musk’s X said it would limit its AI chatbot Grok’s image generation feature to prevent creating images of real people in revealing clothing. The move comes just weeks after the tool faced backlash for producing sexualized images of children and non-consensual nudity.

According to a post on X’s official Safety account, the platform has “introduced technological measures to stop the Grok account from enabling edits to images of real people in revealing clothing, like bikinis.”

The company said this restriction will apply to all Grok users, including paying subscribers.

The company announced it was limiting “image creation and editing” on Grok accounts to paid subscribers. Still, free users could generate images on Grok’s website at the time of publication.

The company also added that it was geoblocking all users from generating “images of real people in bikinis, underwear, and similar attire via the Grok account and in Grok in X in those jurisdictions where it’s illegal.”

The statement did not specify which countries would be impacted by the geoblock.

A look at Grok’s recent timeline revealed at least one case where it followed a user’s request to alter an image of British Prime Minister Keir Starmer, swapping his suit for a bikini. According to The Verge, it’s still “extremely easy to undress women and edit them into sexualized poses” using the X and Grok mobile apps or websites. A UK-based reporter from the outlet also noted she wasn’t prevented from using the app to create “sexualized deepfakes of herself.”

On Wednesday, X owner Elon Musk questioned reports claiming the platform was generating Child Sexual Abuse Material (CSAM), saying he was “not aware of any naked underage images generated by Grok. Literally zero.” He explained that Grok doesn’t create images on its own but only in response to user requests, and it refuses anything illegal, following the laws of each country or state. Musk suggested the problem might stem from users, noting that “adversarial hacking of Grok prompts” could cause unexpected results, which are fixed immediately when discovered.