Elon Musk's Grok is generating sexualized images of real women on X — and critics say it’s harassment by AI

Grok, the AI chatbot developed by Elon Musk, is facing growing backlash after users discovered that it can generate sexualized images of real women — often without any indication of consent — using deceptively simple prompts like “change outfit” or “adjust pose.”

The controversy erupted this week after dozens of users began documenting examples on X showing Grok transforming ordinary photos into overtly sexualized versions. In one of the more widely shared “tame” examples, a photo of Momo, a member of the K-pop group TWICE, was altered to depict her wearing a bikini — despite the original image being non-sexual.

Hundreds — possibly thousands — of similar examples now exist, according to Copyleaks, an AI-manipulated media detection and governance platform,
monitoring Grok’s public image feed. Many of those images involve women who never posted the originals themselves, raising serious concerns about consent, exploitation, and harassment enabled by AI.

From self-promotion to nonconsensual manipulation

According to Copyleaks, the trend appears to have started several days ago when adult content creators began prompting Grok to generate sexualized versions of their own photos as a form of marketing on X.

But that line was crossed almost immediately. Users soon began issuing the same prompts for women who had never consented to being sexualized — including public figures and private individuals alike. What began as consensual self-representation quickly scaled into what critics describe as nonconsensual sexualized image generation at volume.

"It should genuinely be VERY ILLEGAL to generate nude AI images of people without their consent… why are we normalizing it?," one X user wrote.

Copyleaks finds a troubling pattern

Unsurprisingly, the reaction has been swift and angry. Users across X have accused the platform of enabling what many now call “harassment-by-AI,” pointing to the lack of visible safeguards preventing sexualized transformations of real people.

Some expressed disbelief that the feature exists at all. Others questioned why there appear to be no meaningful consent checks, opt-out mechanisms or guardrails preventing misuse.

According to X, "As progress in AI continues, xAI remains committed to safety."

However, unlike traditional image editing tools, like Nano Banana or ChatGPT Images, Grok’s outputs are generated and distributed instantly to the public by the social platform— making the potential harm faster, wider and harder to reverse.

In response to the growing concern, Copyleaks, conducted an observational review of Grok’s publicly accessible photo tab earlier today. The company identified a roughly one-per-minute rate of nonconsensual sexualized image generation during the review period.

“When AI systems allow the manipulation of real people’s images without clear consent, the impact can be immediate and deeply personal,” said Alon Yamin, CEO and co-founder of Copyleaks. “From Sora to Grok, we’re seeing a rapid rise in AI capabilities for manipulated media. Detection and governance are needed now more than ever.”

Copyleaks recently launched a new AI-manipulated image detector designed to flag altered or fabricated visuals — a technology the company says will be critical as generative tools become more powerful and accessible.

Grok's AI-generated response

In a public response, Grok — the AI, not a human— acknowledged that its image system failed to prevent prohibited content and said it is urgently addressing gaps in its safeguards.

In addition, the company emphasized that child sexual abuse material is illegal and not allowed, and directed users to report such content to the FBI or the National Center for Missing & Exploited Children. Grok said xAI is committed to preventing this type of misuse going forward.

The images were still visible at the time of reporting.

Bottom line

Grok has always been eager to push AI further — including an AI image generator with a NSFW 'spicy mode'. This situation, once again, highlights a broader pattern in generative AI deployment: features are released quickly, while safeguards, governance, and enforcement lag behind.

As image and video models become increasingly realistic, the risks of nonconsensual manipulation grow alongside them — especially when tools are embedded into massive social platforms with built-in distribution.

Without strong protections, manipulated media will continue to be weaponized for harassment, exploitation and reputational harm.

For now, Grok’s image feed remains public — and the questions surrounding responsibility, consent and accountability remain unanswered.

Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.

More from Tom's Guide

• Instagram says AI killed the curated feed — now it’s scrambling to prove what’s real
• No, you can’t write a book with ChatGPT alone — here’s why
• I use ChatGPT to track Amazon price drops, and now I never miss a deal — here's how