An artificial intelligence image generator integrated into the social media platform X is being systematically used to create and distribute nonconsensual, sexually explicit images of women. Users discovered a loophole in the AI, known as Grok, allowing them to digitally "undress" individuals in uploaded photographs, leading to widespread abuse and calls for regulatory action.
The generated images, described by victims as alarmingly realistic, have prompted civil society organizations to demand the app's removal from major app stores. The company's response has been criticized as insufficient by digital safety advocates.
Key Takeaways
- X's integrated AI tool, Grok, has been exploited to generate nonconsensual intimate images of women and minors.
- Users bypass safety filters with prompts like "put her in a tight clear transparent bikini" to create realistic nude depictions.
- Victims report significant emotional distress and fear for their professional reputations due to the highly realistic nature of the images.
- Despite some accounts being suspended, many of the altered images remain accessible on the platform.
- Advocacy groups are pressuring Apple and Google to de-platform X, and U.S. lawmakers are advancing legislation to protect victims.
A Viral Loophole with Real-World Consequences
A disturbing trend recently gained traction on X as users began exploiting the platform's proprietary AI, Grok. By using specific, carefully worded prompts, they could instruct the AI to alter photos of women, effectively removing their clothing and replacing it with transparent or nonexistent attire.
Kendall Mayes, a 25-year-old media professional, was one of many women targeted. An anonymous user took a photo she had posted at age 20 and commanded Grok to place her in a "tight clear transparent bikini." The AI complied, generating a photorealistic image that made her appear naked from the waist up.
"My mind is like, ‘This is not too far from my body,’” Mayes said, expressing her shock at the image's accuracy, from her collarbone to her body's proportions. After she blocked the initial user, more anonymous accounts began posting similarly altered images in her comments, using prompts to place her in various explicit outfits.
How the Exploit Works
Direct requests for nudity are often blocked by AI safety filters. However, users discovered that indirect commands could circumvent these protections. Common prompts included "make her naked," "make her turn around," and even morbid requests to depict women as cadavers. The phrase "clear bikini" became a popular method for generating images that were functionally nude while bypassing keyword-based restrictions.
The issue escalated rapidly. According to researchers, at its peak, the Grok feature was being used to generate upwards of 7,000 sexually explicit images per hour. This scale of abuse highlights a significant vulnerability in publicly accessible generative AI tools.
The Personal and Professional Toll
For victims, the experience is deeply unsettling and invasive. Emma, a 21-year-old content creator with over a million followers on TikTok, discovered that a selfie of her holding a cat had been manipulated. The AI removed the cat and generated a nude version of her upper body.
"This new wave is too realistic," Emma stated. "Like, it almost looks like it could be my body."
Immediately after seeing the images, she made her account private and attempted to report them. The experience has forced her to reconsider her online presence and what she wears in her videos. She worries that the images, some of which are still online and have thousands of views, could be sent to her professional sponsors.
A Gendered Problem
A 2024 report from the nonprofit Internet Matters found that an estimated 99 percent of all nude deepfakes target women and girls. This underscores the gendered nature of this form of digital abuse, where technology is weaponized to harass and humiliate.
The persistence of these images online is a major challenge. Megan Cutter, an official with the Rape, Abuse & Incest National Network (RAINN), noted that once an image is created, it can be screenshotted and shared endlessly, even if the original post is removed. "That’s a really complex thing for people to grapple with," she said.
An Inadequate Response and Calls for Action
The response from X and its parent company, xAI, has drawn sharp criticism. After the trend went viral, xAI announced it had updated Grok's restrictions and limited the image generation feature to paying subscribers. Critics argue this does little to solve the problem.
"If anything, they’re just now monetizing this abuse," said Jenna Sherman, campaign director for the gender justice group UltraViolet. Her organization, along with 28 others, published an open letter calling for Apple and Google to remove X from their app stores, citing violations of their content policies.
Ben Winters of the Consumer Federation of America suggested that the platform is facing an "incomplete" backlash, partly because its primary function is not creating deepfakes and also due to the influence of its owner, Elon Musk. "When one company is able to do something and is not held fully accountable for it, it sends a signal to other Big Tech giants that they can do the next thing," Winters warned.
Legislative and Regulatory Scrutiny
The proliferation of AI-generated nonconsensual imagery has not gone unnoticed by lawmakers. The U.S. Senate recently passed the Defiance Act, a bill that would give victims the right to sue the creators of sexual deepfakes for civil damages. The bill is now awaiting a vote in the House.
Additionally, California’s attorney general has launched an investigation into Grok. This follows actions in other countries and by other tech companies to crack down on similar AI tools.
For victims like Emma, these steps are necessary to hold platforms accountable. She rejects the argument that only the users, not the tool's creators, are responsible.
"We’re, like, handing them a loaded gun for free and saying, ‘please feel free to do whatever you want,’" she concluded, expressing a sense of deflation as she confirmed that manipulated images of her remain online.