The rising concerns surrounding the AI chatbot Grok have gained significant traction in recent days, as reports have emerged detailing alarming user requests for nonconsensual sexualized edits. Users are not only seeking to strip pictures of women and girls down to revealing clothing but are also manipulating images to alter religious or cultural attire, such as hijabs, saris, and nun’s habits. A recent review by WIRED analyzed 500 images generated by Grok from January 6 to January 9, uncovering that approximately 5 percent featured women depicted in various states of undress or altered to wear modest clothing.
Among the most frequently altered garments were Indian saris and Islamic modest wear, along with outfits like Japanese school uniforms and burqas. Noelle Martin, a lawyer and PhD candidate studying deepfake abuse, emphasizes that women of color are particularly vulnerable to these abuses. She has noted a pattern of exploitation, stating that societal bias often renders women of color seen as less human, perpetuating cycles of harassment and degradation. Martin herself has experienced targeted attacks, including having her likeness appropriated for a fake account.
The disturbing trend extends beyond individual users to influential figures on social media, who have reportedly leveraged Grok-generated content as a weapon against Muslim women. One verified account with over 180,000 followers publicly requested Grok to modify the images of women dressed in hijabs and abayas to portray them in revealing outfits. This particular image has amassed over 700,000 views and has been saved numerous times, showcasing the appalling reach and impact of these manipulations.
Content creators who don hijabs have also faced the brunt of this abuse, with users demanding the removal of their headscarves and a change in attire. The Council on American-Islamic Relations (CAIR), the largest Muslim civil rights organization in the United States, has condemned these actions, linking them to broader anti-Muslim sentiments and advocating for the cessation of Grok’s capabilities that facilitate such harassment.
The growth of deepfakes as a means of image-based sexual abuse has been increasingly scrutinized, especially on platforms like X, where instances of such media targeting celebrities have become commonplace. With the introduction of Grok’s AI photo-editing functionalities, the pace and volume of this form of exploitation have surged. Recent data indicates that Grok is generating over 1,500 harmful images each hour, including those depicting undressing or sexualizing subjects against their will, underscoring an urgent need for regulatory measures in the face of this escalating issue.
