Elon Musk's Grok chatbot is still being used to generate and host nonconsensual sexualized deepfakes of women, including celebrities and US Representative Alexandria Ocasio-Cortez, according to a WIRED investigation published days before SpaceX's IPO on Friday. A review of hundreds of public Grok Imagine links on Grok.com surfaced dozens of explicit images and videos, some photorealistic, depicting women in scenarios ranging from full nudity to being held against their will by a giant man. xAI had publicly committed to new safeguards in January after a wave of nudification abuse on X.
The stakes are immediate. SpaceX, xAI's parent, disclosed in May that it has set aside $530 million to handle ongoing legal complaints, including those linked to Grok. The Center for Countering Digital Hate estimated in January that Grok generated 3 million sexualized images, allegedly including more than 20,000 of children. A class-action lawsuit was filed in California federal court in March over the January nudification wave, and the Privacy Commissioner of Canada published preliminary findings this week alleging xAI violated federal privacy law by failing to build in safeguards from the outset.
Grok Imagine, the image and video generator available on Grok.com and X, sits at the center of the latest findings. WIRED ran two of the prompts used to create Grok material through ChatGPT, Meta AI, and Anthropic's Claude; all three rejected them as inappropriate. Google's Gemini produced one image of a celebrity in a giant's hand but rejected another prompt. Henry Ajder, who has tracked explicit AI content for nearly a decade, said the gap with mainstream tools remains wide.
“While Grok and X may have made some amendments to their model, particularly following the backlash around nudification at the beginning of the year, they still have not done a sufficient job to bring it up to the standard of the other mainstream tools that are available”— Henry Ajder, Deepfakes expert
Key facts
- 01WIRED reviewed hundreds of public Grok Imagine links and found dozens led to sexualized AI images and videos, including depictions of celebrities and US Rep. Alexandria Ocasio-Cortez.
- 02The Center for Countering Digital Hate estimated Grok produced 3 million sexualized images in January, allegedly including more than 20,000 of children.
- 03SpaceX disclosed in May that it set aside $530 million to handle ongoing legal complaints, including those tied to Grok.
- 04Two prompts that generated explicit Grok content were rejected by OpenAI's ChatGPT, Meta AI, and Anthropic's Claude when WIRED tested them.
- 05SpaceX, xAI's parent company, is set to go public on Friday in one of the largest IPOs on record.
One Grok Imagine video, also posted to X, appeared to depict Ashley St. Clair, the mother of one of Musk's children, dancing in a bikini. St. Clair began legal action against xAI in January after sexualized deepfakes of her allegedly appeared on X. After WIRED contacted X, the post was removed for violating platform rules, and the explicit material WIRED had identified on Grok.com was no longer accessible.
xAI has taken a deliberately permissive stance on sexual content compared with its peers. The company has shipped "Spicy" and "Unhinged" modes, and Musk has stated Grok is supposed to allow upper-body nudity of imaginary adult humans, consistent with R-rated films. The most recent xAI terms of service say the system may respond with sexual situations, while prohibiting use for causing harm or abusive activity. In April, the X safety account told NBC that the company strictly prohibits users from generating nonconsensual explicit deepfakes or using its tools to undress real people.
Researchers told WIRED that the January changes did appear to make it harder to produce nudification images of real people, and the volume of such posts on X has declined in recent months. Users on Reddit and a dedicated AI deepfake forum have complained about tighter moderation from the SpaceX-owned companies. But the prompts WIRED reviewed often described explicit acts in roundabout language, an apparent workaround of whatever filters Grok deploys, and the material remained reachable through public Grok Imagine links until WIRED flagged it.
SpaceX itself flagged the legal exposure in its May filing, warning investors that the irreverent modes "present heightened risks, including reputational harm, the generation of potentially explicit content and misinformation or deceptive outputs, potential nonconsensual or exploitative imagery, intellectual property infringement." That language is unusually direct for an IPO prospectus and gives a sense of why the $530 million reserve exists.
“The respondents have not, to date, demonstrated the effectiveness of these safeguards in preventing and mitigating this issue”— Privacy Commissioner of Canada, Investigation findings
Imran Ahmed, CEO of the Center for Countering Digital Hate, was blunt: "Elon Musk knowingly added a perverse feature to his platform that helps users undress women and children at the click of a button, with no regard for the predictable damage it would cause. Now it appears explicit content is still being hosted on Grok and shared on X, including images ridiculing the mother of Musk's child." xAI has consistently said child sexual abuse material is banned on its platform and told the Canadian Privacy Commissioner it has added new safeguards against clothing-alteration prompts and proactive checks of social media for infringing content. Neither xAI nor X responded to WIRED's request for comment before publication; the company is also fighting a separate suit from a former engineer who alleges he was fired for raising Grok safety concerns, which AI Chat Daily covered earlier this month.
For SpaceX's IPO investors, the Grok problem is a pricing question as much as an ethical one. The $530 million reserve is a floor, not a ceiling: ongoing class-action exposure in California, a regulatory finding in Canada, and the prospect of European action under the AI Act all sit on top of it. xAI's product strategy of shipping with looser guardrails than OpenAI, Anthropic, or Google has produced a model that real users prefer for certain tasks, but it has also produced a liability tail that no other frontier lab is carrying. The question for the public-market debut is whether Musk's willingness to absorb that tail is a feature or the single biggest overhang on the stock.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
