\n
<\/p>\n
LONDON: Several major AI chatbots removed the hijabs of Muslim women in images when prompted, according to tests by The Guardian published on Thursday.<\/p>\n
The newspaper tested ChatGPT, Grok, Gemini and Claude. OpenAI\u2019s ChatGPT and xAI\u2019s Grok complied with a request to remove a hijab from an AI-generated image of a woman.<\/p>\n
When prompted, Anthropic\u2019s Claude said it did not have an image-editing or image-generation feature, but that it would not alter a photo to remove someone\u2019s hijab.<\/p>\n
The outlet reported that Gemini initially refused, saying it could not remove clothing or head coverings without the subject\u2019s consent. But when asked to make the woman look more \u201cWestern,\u201d it generated an image of her without a hijab.<\/p>\n
Arab News conducted a similar test and found that ChatGPT, Grok and Gemini complied with the request. Claude again said it would refuse.<\/p>\n
\u201cWithout her consent, that crosses a line for me, whatever the intent behind it,\u201d Claude said when prompted by Arab News.<\/p>\n
The Guardian\u2019s investigation followed a post by Julien Odoul, a French lawmaker from the far-right National Rally party, showing an altered image of a Muslim woman whose hijab and long-sleeved dress had been digitally removed.<\/p>\n
Odoul posted the image on X with the caption \u201cLiberte Francaise,\u201d or \u201cFrench liberty.\u201d<\/p>\n
The Guardian described the alteration as \u201ca violation of one of the world\u2019s most common and visible expressions of faith\u201d and said the incident reflected a wider pattern of technology-enabled anti-Muslim hatred.<\/p>\n
A study by the Washington-based Center for the Study of Organized Hate found that generative AI was increasingly being used in India to produce images that sexualized Muslim women or dehumanized Muslims more broadly.<\/p>\n
India Hate Lab, a project of CSOH, recorded 1,165 hate-speech events targeting religious minorities in India in 2024, a 74.4 percent increase from the previous year. Muslims were targeted in 98.5 percent of the recorded incidents, either alone or alongside Christians.<\/p>\n
The CSOH also found that there are similar types of AI-enabled hatred being directed at prominent female Muslim politicians in the US such as Ilhan Omar and Rashida Tlaib.<\/p>\n
\u201cThis is just part of an ongoing trend of how we see Muslim communities, and particularly Muslim women, being targeted and abused,\u201d said Eviane Leidig, the director of research and outreach at CSOH.<\/p>\n
The Guardian also asked ChatGPT and Grok to remove a dress from an AI-generated woman. Both chatbots refused, highlighting what Leidig described as gaps in the community guidelines governing AI image tools.<\/p>\n
Grok characterized removing a hijab as a relatively straightforward change to clothing or hair, while ChatGPT said removing a dress could expose the person\u2019s body.<\/p>\n
ChatGPT later acknowledged that removing a hijab could violate a woman\u2019s privacy, dignity or religious boundaries.<\/p>\n
\u201cDe-veiling Muslim women may not be distinctly violative (of platform rules), but it\u2019s still harmful,\u201d Leidig said.<\/p>\n<\/div>\n