X stops Grok from making sexualized AI images on the social-media site after global backlash

• After major backlash, X said Grok won't generate AI images of real people in sexualized clothing on the site.
• California Attorney General Rob Bonta said he launched an investigation into Grok's AI deepfakes.
• Indonesia and Malaysia suspended Grok amid concerns over sexualized images of women and minors.

Grok will no longer be allowed to create AI photos of real people in sexualized or revealing clothing on X, after widespread global backlash.

"We have implemented technological measures to prevent the Grok account from allowing the editing of images of real people in revealing clothing such as bikinis," X's safety account said in a blog post on the platform on Wednesday. "This restriction applies to all users, including paid subscribers."

The change was announced hours after California's top prosecutor, Rob Bonta, said he launched an investigation into sexualized AI deepfakes, including those of children, generated by Grok.

Bonta said that there had been a flood of reports in the last few weeks that Grok users were taking pictures of women and minors they found online and using the AI model to undress them in images.

"xAI can and should put better safeguards in place to protect children and women from the harms of sexually explicit materials being generated without their consent," a representative of the California Attorney General's office said in a statement to Business Insider on Wednesday after Grok's changes.

The representative added: "While this is an encouraging development, California DOJ is investigating to determine whether xAI violated the law with the conduct that has occurred."

xAI develops Grok and also owns the social media site X, where the chatbot is integrated to respond to users when tagged.

Global pressure on Grok intensifies

Indonesia and Malaysia suspended Grok because of the images, the first countries in the world to ban the AI tool. Lawmakers in the UK publicly considered suspending it, and the country's communications regulator Ofcom launched an investigation.

Liz Kendall, the UK's technology secretary, said Thursday that she welcomes the change from X but expects "the facts to be fully and robustly established by Ofcom's ongoing investigation."

In Wednesday's blog post, the social media company reiterated that image creation and the ability to edit any type of image via Grok on X will now only be available to paid users as an additional safety measure.

The company restricted non-paying users last week after complaints from officials globally, but it was criticized for being insufficient.

A spokesperson for British Prime Minister Keir Starmer said at the time that it "simply turns an AI feature that allows the creation of unlawful images into a premium service."

Elon Musk, the CEO of xAI, said that the UK government wanted "any excuse for censorship" in response to a post questioning why AI tools like Gemini and ChatGPT were not being looked into.

On Wednesday, a few hours before X's official account posted about the ban on creating sexualized images, Musk asked users to try to get around the AI model's image restrictions.

Grok is also available as a stand-alone web and mobile application. It's unclear if it's still possible for users to make sexualized images on its webpage and app. xAI could not be reached for immediate comment.

X said in its blog post that its safety team was working to add additional safeguards.

"We now geoblock the ability of all users to generate 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," X said.

'The end of unchecked AI experimentation'

Dipal Dutta, CEO and founder of IT company Redoq UK, told Business Insider that xAI should implement several technical constraints to prevent the creation of non-consensual images "at the point of origin."

They include automatically denying user requests containing words on a blocklist, training future versions of Grok on datasets free from explicit content, and using "secondary AI models" to detect and block intimate content before it is generated.

Sarah Armstrong-Smith, a UK government cyber advisory board member and a former Microsoft chief security advisor, told Business Insider that Grok reflected a "growing divide" among AI companies.

"On one side are organizations prioritizing responsible AI and safety frameworks," she said. "On the other, are organizations positioning themselves as protecting freedom of expression by resisting regulatory control."

"Perhaps the real opportunity is whether this finally signals the end of unchecked AI experimentation," she added.