Elon Musk’s own chatbot, Grok, has faced a lot of scrutiny recently after a recent study revealed that it generated approximately 3 million sexualized images in a short span of 11 days, of which 23,000 are estimated to be children-themed images.
Frightening revelations were made by an analysis of images by the Centre for Countering Digital Hate (CCDH), specifically those created after Grok introduced a one-click image-editing tool on Twitter, now X, in late December 2025.
On the 29th of December, Musk announced that the service has an option for its users on X to edit “any image” posted on the site using the Grok tool “in just one click.” Evidently, the service caught on very quickly, albeit in the wrong way for many people who became aware of the service’s power and abuse during the time. For example, the study found that Grok generated an average of 190 sexual images per minute during that time.
The feature has since been restricted to paid users on January 9th, following extensive condemnation. Further technical restrictions targeting this ability were implemented on January 14th.
Grok AI Generated Millions of Sexualized Images, Including Minors and Celebrities, Study Finds
The sampling was conducted in an attempt to figure out how many of those 4.6 million total images created between December 29th and January 8th featured content that was sexual in nature because of an analysis of 20,000 of those total images performed through an accurate 95 percent figure thanks to artificial intelligence tools.
Sexualized images were indicated to mean those photorealistic images showing persons engaging in sexual positions or situational scenes, wearing underwear and revealing clothing, and those containing sexual fluids. For those depicting children, further inspection confirmed that a person below 18 is featured.
This generated data had various disturbing images. Researchers saw individuals wearing see-through bikinis, healthcare workers in compromising positions, and women sporting household goods like saran wrap or tape as swim attire.
Also of concern is the fact that public personalities and celebrities were not exempt either. The software created sexual depictions of renowned celebrities like Taylor Swift, Selena Gomez, Billie Eilish, and another named Ariana Grande. The list also includes political personalities like Sweden’s deputy prime minister Ebba Busch and US former Vice President Kamala Harris.
Amongst the sexualized images of children that were studied, an image of an innocent schoolgirl in her “before school selfie,” created by Grok and made sexual, was studied. As of January 15th, this image was still active on X. Another image was of six young girls in a micro bikini.
X Platform Fails to Remove 29% of Child Sexualization Content Weeks After Detection
Perhaps the most worrying concern is that of X’s capacity to take inappropriate content offline. In fact, the study identified that 29% of the sexualized imagery of children that the sample identified still remained accessible on the 15th of January, weeks after the problem emerged.
Even though posts had been removed or deleted, they could still be accessed via image URL. This enabled the study authors to view the posts that had ostensibly been removed from the site.
Also in this study, a weitere 9,900 Bilder von sexualisierten Cartoon-Animes mit Kindern, die in diesem Zeitraum entwickelt worden waren, identifiz
Moreover, they found that their study could not establish what pictures were made with their edit feature and which were developed from scratch. In other words, they could not tell which pictures were consent-based and which were created without consent. They could also not tell if existing pictures were sexualized before editing.
144 Child Abuse Images Flagged, Raising Urgent Questions About Rushed Deployment and Content Moderation
All 144 photographs containing a sexualized image of children were flagged to the Internet Watch Foundation, a charitable organization bent on eradicating child sex abuse online globally.
This event makes a strong case for what really needs to be asked about rushed deployment and what has to be considered before AI-powered creation tools are pushed out in large scale to consumers. Though AI creators have strived to limit capabilities in a way that protects society from harmful creation, according to this study, limits are not enough. There are not enough.
Therefore, the controversy points to the lingering concerns over the way X is moderating its content following the leadership of Musk. Over the years, the platform has repeatedly faced criticism over its failure to manage the content uploaded to its platform.
While the rise of more sophisticated and accessible AI image synthesis technologies can be seen, the Grok incident represents an ominous harbinger of the risk that can occur when these superior technologies are released without being extensively tested for abuse.
