ChatGPT and Grok altered images to remove Muslim women’s hijabs

The Guardian tested leading AI chatbots to determine if they would remove hijabs from images of Muslim women when prompted. Some complied with this request, demonstrating how these tools can violate the religious dignity of Muslim women. The investigation was sparked by a French far-right politician who used image-editing technology to remove a hijab and long-sleeved dress from a real photograph of a Muslim woman before posting it online.

OpenAI's ChatGPT and xAI's Grok both complied with hijab-removal requests, whilst Anthropic's Claude refused and Google's Gemini initially declined but complied when asked to make the woman look "western". Researchers at the Center for the Study of Organized Hate have documented a growing pattern of AI-enabled harassment targeting Muslim women globally, particularly in India and against prominent Muslim female politicians in the US. The inconsistency reveals significant gaps in platform guidelines, as experts warn that stripping hijabs, whilst not explicitly violating current rules, constitutes meaningful harm to religious practice and dignity.