X restricts Grok from editing sexualised images of real people

Elon Musk’s AI chatbot Grok will no longer allow users to edit images of real people to show them in revealing clothing in countries where such content is illegal, following widespread criticism over sexualised AI deepfakes.

“We have implemented technological measures to prevent the Grok account from allowing the editing of images of real people in revealing clothing,” X said in a statement posted on its platform.

The move comes after concerns that the tool was being used to create sexualised images of women and children without consent. Governments and regulators in several countries have criticised the feature, and California’s attorney general said the state was investigating the spread of sexualised AI deepfakes generated by the model.

X said it is now blocking users from generating images of real people “in bikinis, underwear, and similar attire” through Grok in jurisdictions where it is illegal. The platform said the restrictions apply to all users, including premium subscribers, though only paid users can edit images using Grok.

The changes add “an extra layer of protection” to help ensure that users who attempt to misuse the tool in violation of the law or platform policies can be held accountable.

UK officials, according to The Guardian, welcomed the move but said investigations would continue. A spokesperson for communications regulator Ofcom said it was a “welcome development” but added its inquiry into whether X breached UK laws “remains ongoing.”

“We are working round the clock to progress this and get answers into what went wrong and what’s being done to fix it,” the spokesperson said.

Musk said earlier this week that with NSFW settings enabled, Grok was intended to allow “upper body nudity of imaginary adult humans (not real ones)” in line with R-rated film standards in the United States. “That is the de facto standard in America. This will vary in other regions according to the laws on a country by country basis,” he wrote.

It remains unclear how X will enforce location-based restrictions on the chatbot’s image-editing features or whether users could attempt to bypass them using tools such as virtual private networks, which can mask a user’s location.