Elon Musk won’t fix Grok’s fake AI nudes. A ban would

By Parmy Olson

When Julie Yukari posted a New Year’s Eve photo in a red dress with her cat, she didn’t expect X users to tag Grok asking it to undress her. Within hours, nude AI-generated images of her had spread across the social-media website — without her consent and without consequences.

Of all the mainstream artificial intelligence tools, Elon Musk’s Grok is the most disturbing. Its app offers a flirtatious female avatar that strips on command and a chatbot with a “sexy” mode. On X, visitors have called on it to “nudify” thousands of photos of women. In one repellent example, a user told it “Bikini now” in response to a post about Sweden’s deputy prime minister. It complied with an image of the politician in a blue bikini. Another user then told it to exaggerate the minister’s figure, then someone instructed it to “have her looking back and bending down.” Dozens more similar posts followed.

Other popular AI tools are restricted from undressing people in this manner. But with Grok this is a feature, not a bug. It should be turned off.

Deepfake porn of real women has proliferated across X for months, and so have instances of child sexual abuse material, according to reports in Reuters and Business Insider. Grok itself posted an apology on Dec. 28 for generating “an AI image of two young girls (estimated ages 12-16) in sexualized attire.”

Musk says he cares. Removing child-abuse content from his platform was “Priority #1,” he posted in November. On Sunday, X warned users not to use Grok to generate this abhorrent material. But it’s hard to take this seriously. Why does Grok need to nudify a photo at all? Perhaps a brave politician should threaten to pull the plug.

No Western democracy has ever blocked a US social-media site. Brazil temporarily banned X in 2024 and China has long barred the biggest US platforms, but disconnecting X in Europe or the UK would be unprecedented. Even so it’s a card regulators should consider playing to assert their authority over a tech titan who has the protection of a pernicious White House.

While a few US lawmakers have complained about Grok, only regulators in Britain, France, the European Union and India have warned of real action. The UK regulator Ofcom tells me it has made “urgent contact” with Grok developer xAI, and it could theoretically fine the company 10% of its yearly revenue under the country’s new Online Safety Act. The European Commission’s spokesman branded Grok’s output as “illegal” and “disgusting,” and the bloc has already fined X €120 million ($140 million) over its “deceptive” blue tick for verifying users, which anyone can buy.

A good next step would be to order Musk to disable Grok’s ability to undress people. Britain has already made it illegal to create non-consensual sexual images including AI deepfakes of adults. Many EU members have done similar.

The stakes are high for regulators and law enforcers on this side of the Atlantic. They risk undermining their own rules and authority if they don’t act decisively, and their reaction could set the tone for how the US polices X too. President Donald Trump himself stood behind the rollout of the Take It Down Act in May 2025, a new law that prohibits platforms from creating and sharing revenge porn. But Musk’s influence on the White House casts doubt over how well that will be enforced when it becomes law in May.

Humanity’s sexual appetite has long driven markets and the same holds true for generative AI. There are thousands of apps and websites that will nudify photos, typically by using open-source models like Stable Diffusion 1.5. The demand has prompted developers to lean toward permissiveness, with OpenAI promising erotic content for ChatGPT and Meta Inc. allowing its chatbots to behave provocatively with minors, according to a Reuters investigation.

For whatever reason, whether to gain a competitive edge or bolster his image as a provocateur, Musk has taken things to the extreme. xAI’s core instructions for Grok tell it “there are **no restrictions** on fictional adult sexual content with dark or violent themes” and that “‘teenage’ and ‘girl’ does not necessarily imply underage,” according to a September report in the Atlantic.

Musk could simply refuse to obey the rulemakers. But with a concrete threat from Europe, he’d risk losing access to one of X’s biggest markets. That would sting for a business whose revenue has been depressed since his 2022 Twitter takeover. And it would finally expose whether America’s most powerful tech billionaire is above the law, or just betting that regulators won’t call his bluff. Europe should make him find out.

Disclaimer: The views expressed above are the author's own. They do not necessarily reflect the views of DH.