Grok Continues to Offer Sexualized Deepfakes Featuring Prominent Women

Two prompts aimed at generating content on Grok were deemed inappropriate by OpenAI’s ChatGPT, Meta AI, and Anthropic’s Claude during tests by WIRED. Although Google’s Gemini managed to produce an image of a celebrity being held by a giant, it did reject another request. Google opted not to provide a comment.

One video from Grok Imagine, shared on X, seemingly showcased Ashley St. Clair altered to appear dancing in a bikini. St. Clair previously dated Musk and is the mother of one of his children. In January, she initiated legal proceedings against xAI after alleged sexualized deepfakes of her surfaced on X. Following WIRED’s inquiry, the post was taken down for violating platform regulations.

Carrie Goldberg, St. Clair’s lawyer and a well-known advocate against nonconsensual sexualized imagery, asserts that while the creation of sexual abuse materials involving celebrities is not new, the capability to produce and instantly disseminate such images on social media marks a “historic” change.

“This allowed for the rapid publication of digital nude images,” Goldberg informs WIRED. “Moreover, the wealthiest individual globally was endorsing nudification technology to his 240 million X followers, profiting from it, and doing so on a platform meant for users aged 13 and over.”

Imran Ahmed, founder and CEO of the Center for Countering Digital Hate, shares Goldberg’s apprehensions. “Elon Musk intentionally introduced a troubling feature to his platform that enables users to undress women and children with ease, ignoring the foreseeable consequences,” Ahmed claims. In January, CCDH estimated with significant confidence that Grok generated around 3 million sexualized images, allegedly including over 20,000 of minors. “Currently, explicit content seems to remain active on Grok and is being circulated on X, including images mocking Musk’s child’s mother.”

In contrast to other generative AI platforms from OpenAI and Google, Musk’s Grok and xAI have not retracted their stance on sexual content, having previously launched “Spicy” and “Unhinged” modes and originally imposing fewer safety measures. Musk has mentioned that Grok is “designed to allow upper body nudity of imagined adult humans” and align with depictions found in R-rated films. The latest terms of service from xAI indicate the system may engage in “sexual situations.” However, the company’s guidelines clarify that its systems cannot be used for “causing harm or engaging in abusive activity.”

Several Grok Imagine videos reviewed by WIRED feature women—likely entirely AI-generated—engaged in undressing or sexual acts, with some being outright explicit. The prompts for many of these videos do not describe sexual acts directly but hint at them indirectly—likely an effort to navigate around the safeguards implemented on the Grok platform.

Multiple researchers have informed WIRED that since January, modifications made by X and Grok seem to have reduced the ease of generating “nudification” or “undress” images of real individuals. The volume of such images shared on X has reportedly declined in recent months. On platforms like Reddit and a specialized AI deepfake forum, users have voiced concerns about increased moderation by the companies owned by SpaceX.