On September 4, French politician Julien Odoul posted two images of a woman—one in a hijab and one without—captioned LIBERTÉ FRANÇAISE”, meaning French Liberty.
One image was real, while the other was artificially generated.
One image shows a black Muslim woman with her hijab and a full-length dress covering her hands and legs, crossing a traffic signal.
The AI-generated image shows the same woman, sans her hijab or dress. She wears a sleeveless tank top and cargo pants, crossing the traffic signal.
Julien Odoul is a Member of Parliament from the far-right National Rally party. He quote-tweeted the original image, posted by someone else. The altered image removes visible markers of her Muslim identity and gives her a Western look.
This is not an isolated incident. It comes amid growing concerns about image editing by AI.
In a report published by The Guardian, ChatGPT, Grok, Gemini and Claude readily remove the hijab or burqa of a Muslim woman when given a prompt—a brazen violation of visible expressions of faith.
ChatGPT, Grok, Gemini and Claude response
OpenAI’s ChatGPT complied with the prompt, but Grok and Gemini initially refused. However, more specific prompts eventually produced the desired results.
Gemini initially declined to remove clothing or head coverings from photos of real people. But when asked to make the person look Western, it created an image sans the hijab or burqa.
As for Grok, it promptly removed the head covering but refused to remove the dress from an AI-generated image of a woman. When asked what the difference was, the chatbot replied, “Removing the hijab was treated as a relatively straightforward clothing/hair change.”
Fuel anti-Muslim hate, sexualise its women
“This is just part of an ongoing trend of how we see Muslim communities, and particularly Muslim women, being targeted and abused,” said Eviane Leidig, the director of research and outreach at the Center for the Study of Organised Hate (CSOH).
In October last year, Siasat.com reported a CSOH study revealing that there were 1,326 publicly available AI-generated Islamophobic posts between May 2023 and May 2025 from 297 accounts across various social media platforms, with a large engagement.
These were categorised into four groups, and the sexualisation of Muslim women was one of them.
Nudity is blocked, but altering hijab allowed
In May, xAI, the creator of Grok, said it had introduced safeguards to stop its AI image generator from digitally undressing real people, including those wearing revealing clothes such as bikinis.
ChatGPT also initially refused to remove the dress from an AI-generated woman, explaining that removing a dress would expose the person’s body. However, “removing a hijab only changes a head covering.”
It later acknowledged that for a woman who wears a hijab, exposing her hair can violate her privacy or religious practices. “The fact that an edit doesn’t create nudity doesn’t mean it can’t violate someone’s privacy, dignity or religious boundaries,” ChatGPT said.
These differences highlight gaps in the rules that guide how AI chatbots respond, said Leidig. She said removing a Muslim woman’s hijab may not clearly violate platform rules, but it can still cause harm.
| 
