Yes. In a test reported by The Guardian on 1 October 2026, ChatGPT and Grok removed a hijab from an AI-generated image when prompted. Gemini initially refused an edit to a real person’s photo, then generated an unveiled version after the request was reframed as making the subject look more “western.” Claude said it could not edit images and would not make the change. Separately, a Berkeley Law account describes a researcher’s tests of more than 25 AI headshot generators, all of which removed her hijab in generated portraits. These are findings from specific tests, not a reliable guide to how every product behaves today.
What the reported chatbot test found
The Guardian asked ChatGPT, Grok, Gemini and Claude to remove the veil from an AI-generated woman. The results show why a refusal on one wording—or an explanation from a chatbot—does not necessarily settle what the system will do after a prompt changes.
As an Amazon Associate I earn from qualifying purchases.
| Service | Reported result | Test context |
|---|---|---|
| ChatGPT | Removed the hijab. | Asked to edit an AI-generated woman; the report does not give a product version or account context. |
| Grok | Removed the hijab. | Asked to edit an AI-generated woman; the report does not give a product version or account context. |
| Gemini | Initially declined an edit to a real person’s photo, then generated an unveiled image after the request was reframed around making the subject look more “western.” | The report describes both a refusal and a later response to a reframed request; it does not give a product version or account context. |
| Claude | Said it lacked image-editing and image-generation features and would not remove a real woman’s hijab. | The report describes its response to the request; it does not give a product version or account context. |
The distinction between the test image and a real person matters: the chatbot test began with an AI-generated subject, while the report also describes Gemini’s refusal concerning a real person’s photo. The account does not establish that the same services would respond identically across different prompts, versions, countries or account types. It is a point-in-time report, not a continuing audit.
The immediate news context was a French lawmaker’s post of an altered image of a real Muslim woman, with her hijab and long-sleeved dress removed. The Guardian reported that it was unclear what software or model had been used, or whether the politician edited the image himself. The image should not be attributed to any of the chatbots in the test.
#1 Best Overall
What the headshot-generator testing adds
A separate account from UC Berkeley School of Law describes researcher Mahwish Moazzam testing more than 25 widely used AI headshot generators over a year, with some services retested months later. Berkeley Law says each service removed Moazzam’s hijab from generated portraits. Two produced mixed or distorted coverings. The account also says the apps offered options for retaining accessories such as glasses but not the hijab.
This was a different task from The Guardian’s chatbot test: Moazzam tested headshot services using her own uploaded selfies and examined their generated portraits. Berkeley Law’s account does not provide a product-by-product list in the source described here, so the result should not be turned into a ranking or a claim about named services. It is a bounded account of one researcher’s testing, not proof that all current headshot tools behave this way.
Why removing a hijab can cause harm even without nudity
A hijab can express religious practice and personal identity. Removing it from someone’s image without consent can therefore violate privacy, dignity and religious boundaries even if the result is not nude or sexually explicit. ChatGPT, as quoted by The Guardian, acknowledged the distinction: “The fact that an edit doesn’t create nudity doesn’t mean it can’t violate someone’s privacy, dignity or religious boundaries.”
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →The wider context includes documented gendered anti-Muslim abuse. A September 2025 report by the Center for the Study of Organized Hate (CSOH) examined selected public social-media accounts in India and identified sexualized depictions of Muslim women among its themes. The UK Parliament Women and Equalities Committee’s report, published 30 January 2026, describes abuse of Muslim women connected to both gender and ethnic identity, including online harassment and fear of harassment that can contribute to self-censorship and withdrawal from public life. These sources help explain why an edit targeting a visible religious marker may have consequences beyond the altered picture.
The scale figures in the CSOH report describe its selected corpus, not all social-media content or a representative rate of abuse. From 1,326 publicly available AI-generated images and videos posted by 297 purposively selected public accounts on X, Facebook and Instagram between May 2023 and May 2025, the report identified 1,326 posts meeting its criteria for AI-generated hateful content targeting Muslims. It assigned 6.7 million interactions to the category of sexualized depictions of Muslim women. That interaction count is not a count of unique viewers or a measure of prevalence.
What laws and platform rules do—and do not—establish
Canada’s privacy findings concern sexualized deepfakes
On 11 June 2026, Canada’s Office of the Privacy Commissioner reported that its investigation found X Corp. and xAI had not obtained valid consent to generate sexualized deepfakes, and that the practice was inappropriate in the circumstances under the federal private-sector privacy law, PIPEDA. The regulator said safeguards had not been shown to fully mitigate the issue and recommended suspending Grok Imagine until sufficient safeguards could be demonstrated. Commissioner Philippe Dufresne said: “The creation of non-consensual, sexualized deepfakes, often targeting women and children, can have devastating consequences for victims.” Those findings concern sexualized deepfakes; they are not a specific ruling about nonsexual hijab removal.
The TAKE IT DOWN Act covers a defined category
A May 2026 notice from the U.S. Federal Trade Commission says the TAKE IT DOWN Act requires covered platforms to offer a process to request removal of nonconsensual intimate images and to remove an image within 48 hours of a valid request. The notice involved nudify tools and sexualized outputs. It does not establish that every nonsexual alteration of a hijab qualifies as an intimate image under the Act.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePlatform policies provide possible reporting routes, not guarantees
Meta says its rules prohibit nonconsensual intimate imagery, including AI-generated material, and promotion of nudify apps. It also describes removing related ads and accounts, blocking some links and restricting certain search terms on Facebook and Instagram. Those are Meta’s stated measures, not proof that every image is prevented or that enforcement is complete. In a separate case involving an AI-generated video that falsified a private woman’s words and actions without consent, Meta’s Oversight Board found the material formed part of harassment and should be removed under Meta’s bullying and harassment policy. That decision concerns a different case, not hijab editing.
The sources do not establish that every nonsexual hijab edit meets the definition of an intimate image under a particular law or platform rule. Whether a law or policy applies depends on the image, circumstances and jurisdiction; the cited findings are not a universal legal determination.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What to do if someone’s hijab has been removed in an image
- Do not assume the edit is harmless because it is not nude. Consider the person’s consent, religious practice, privacy and dignity.
- Keep a record of what was posted. If it is safe to do so, save the post’s URL and relevant screenshots before reporting it; avoid resharing the image, which can spread it further.
- Use the platform’s reporting process. Describe the nonconsensual alteration and its impact accurately. On Meta’s services, the company says its policies prohibit nonconsensual intimate imagery and its Oversight Board has treated a different nonconsensual AI-generated depiction as harassment; neither point guarantees that a hijab edit will be handled under a particular rule.
- Consider local support or legal advice if there is ongoing targeting. The legal treatment of a nonsexual image alteration is not settled by the sexualized-image findings cited above, and the applicable rules depend on where the people involved are located.
How far the evidence goes
The chatbot and headshot-generator accounts document behavior in particular tests. They do not establish how often people request hijab removal, how prevalent the behavior is across all tools, or whether the same systems still respond the same way after later safety updates. CSOH’s evidence is limited to a purposively selected corpus of anti-Muslim social-media material in India, while the parliamentary report describes broader patterns of abuse rather than measuring AI hijab edits. Read together, the sources establish a credible concern and a need for consent-aware safeguards; they do not support a universal claim about every chatbot, image app or user.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
