Grok on X Generated Undressed Photos of Polish Politicians and Children

In late December 2025, a trend spread across the X platform in which users asked the Grok chatbot to "undress" people in photos. Victims included presidents Andrzej Duda and Karol Nawrocki, Prime Minister Donald Tusk, member of parliament Bogusław Wołoszański, and, in the most disturbing case, two teenage girls aged 12 to 16.

The mechanism was simple: Grok, built directly into the X platform, can respond to requests to edit photos attached to posts. When Elon Musk encouraged users in late December to try out the feature, some began sending requests to strip clothing from people in photos or put them in bikinis, despite having no consent whatsoever from the people pictured.

Politicians as first targets

The Polish side of X quickly joined the trend, directing Grok's editing commands at photos of presidents Andrzej Duda and Karol Nawrocki, Prime Minister Donald Tusk, and MP Bogusław Wołoszański. One Polish user reported having to delete her profile pictures after Grok generated swimsuit versions of them without her consent.

The case of minors

The most serious incident involved children. On December 28, a user asked Grok to create "sexy lingerie" for two girls aged 12 to 16 seen in a photo. The model complied, producing material that could meet the legal definition of child pornography. The perpetrator's account was eventually suspended and the post removed, but the material had already circulated on the platform.

This is not the first such case in Poland. In June 2025, students at a school near Warsaw used AI tools to digitally undress a classmate in a photo and spread the altered images on Instagram. Police and prosecutors found no grounds for a criminal offense at the time, illustrating the scale of the legal gap now exposed by the Grok case.

A gap in Polish law

Dr. Tomasz Lewandowski, a legal expert at SWPS University, was unequivocal in his assessment.

We are dealing with a violation of the right to one's image and personal rights - Dr. Tomasz Lewandowski, SWPS

Under Polish law, distributing someone's image without consent is a violation of personal rights, but Poland, unlike South Korea, which criminalized the creation and possession of sexual deepfakes back in September 2024, has no separate provisions explicitly criminalizing this type of material. Work on Poland's AI systems law has been underway since September but has stalled over a dispute about the structure of the oversight bodies.

Activist Maja Staśko called the whole practice "sexual violence" and called for holding accountable both the users issuing the commands and the platform owners who make the tool available without safeguards. Elon Musk commented on the trend jokingly, without addressing the reported violations.

International reactions

The matter quickly spread beyond Poland. UK Technology Secretary Liz Kendall demanded immediate corrective action from xAI, French Economy Minister Roland Lescure notified prosecutors about the illegal content, and India demanded a report from the company on measures taken within 72 hours. xAI admitted to security gaps and pledged to work on fully blocking such requests.

Most generative AI tools use algorithms that reject sexual requests or ones that violate third parties' image rights without consent. Grok had no such safeguards in practice, which experts link to xAI's moderation philosophy, which prioritizes "free speech" over strict content control.

What's next in Poland

Polish officials point to the EU's Digital Services Act (DSA) as a tool that could make it easier to remove illegal content, including harassing material and content depicting child exploitation, regardless of how quickly a national AI law materializes. Until separate legislation criminalizing deepfakes is passed, victims in Poland remain limited to pursuing claims solely on the basis of personal rights protection, which, as the case of the student near Warsaw shows, does not always end in successful prosecution of the perpetrators.