Weaponized AI Targets Muslim Women in India

New Delhi, India – When Samreen Ayoub first saw the video, she was shocked. The freelance model from Indian-administered Kashmir was scrolling through her phone last year when a friend sent her a clip going viral on Instagram. The video appeared to narrate her life story in New Delhi, with voiceover, rolling subtitles and a title mimicking a television news bulletin. But it was entirely fabricated.

“It felt like being actually stalked,” said Ayoub, 24. “They traced my life from the first semester to the last semester at university.” The video spliced photos from Ayoub’s time as a student at Jamia Millia Islamia University in New Delhi – images from everyday campus moments, including group projects, farewell parties and selfies with classmates. The voiceover, generated by artificial intelligence, falsely claimed she was a Muslim woman “selling her body” to Hindu men. It misidentified people in the photos and called her own brother a “pimp.”

“It looked so real that if someone, even my parents, watched the video, they would think it was real,” Ayoub said. She is one of many Muslim women who have experienced what researchers describe as an increasingly clear pattern: the use of AI to create sexualized images and propaganda. Al Jazeera contacted several targeted Muslim women. They declined to speak publicly citing shame and the risk of re-traumatization.

The trend of sexualizing images and videos of Muslim women coincides with India’s growing involvement in global discussions on AI governance, including the high-level AI Impact Summit held in New Delhi earlier this year. A study by the Washington, DC-based Center for the Study of Organized Hate (CSOH) analyzed 1,326 AI-generated images and videos collected from 297 public accounts on X, Facebook and Instagram from May 2023 to May 2025. Researchers found that sexualized depictions of Muslim women generated the highest engagement – over 6.7 million interactions across platforms.

“Generative AI has helped turn sexual fantasies into images quickly and cheaply,” said Zenith Khan, co-author of the study and digital research analyst at CSOH. “Image generators and deepfakes allow individuals to convert hate narratives into extremely realistic visual material with minimal technical expertise.”

Researchers are not the only ones tracking this trend. Meri Trustline, an online safety helpline run by the Mumbai-based RATI Foundation, has also seen a rise in such cases. The helpline’s 2024 report showed a worrying pattern: while the media often focuses on celebrities and politicians, non-public figures are also being targeted with artificially created but real-harm-causing images. Frontline counselors at the helpline, including Salman Mujawar, said they have recorded an increase in such cases. Since its launch in 2022, Meri Trustline has handled more than 482 cases, about 10% of which involve digitally manipulated material – a share that is rising as AI tools become more accessible.

Ayoub’s video spread across multiple social media accounts within hours. Abusive comments, threatening phone calls and accusations about her character quickly followed. “It felt like a digital lynch mob,” she said. “Not one but more than a dozen accounts posted that video everywhere, and hundreds of others were sharing it.”

The CSOH dataset includes AI-generated memes depicting Muslim women in religious attire in sexually suggestive situations as well as fabricated pornographic images targeting journalists and activists. Researchers observed a recurring visual pattern: a ‘code Muslim woman’ accompanied by a ‘code Hindu man’. “In these narratives, Muslim men are often portrayed as violent or morally corrupt,” Khan said. “Meanwhile, Muslim women are portrayed as submissive or as being ‘saved’ by men from the majority community.”

This imagery, researchers argue, is not accidental in political discourse – it is part of it. Sahana Udupa, a media anthropologist at Ludwig Maximilian University of Munich, described the phenomenon as part of a broader “politico-pornography” targeting women and minority communities. Right-wing digital cultures, she said, combine humor, memes and sexualized imagery to normalize abuse. That ecosystem, scholars note, has deeper ideological roots than mere misogyny. Writing in the South Asia Multidisciplinary Academic Journal, researcher Soma Basu argued that what is happening is the politicization of sexuality itself. The Muslim female body has become a battleground for community dominance – a dynamic that found its clearest expression in the ‘Sulli Deals’ and ‘Bulli Bai’ controversies, fake auction platforms targeting Muslim women in India.

“In many South Asian cultures, women are portrayed as the honor of the family,” Khan said. “So attacking Muslim women with images becomes a way to portray Muslims as inferior.” In response to these concerns, BJP politician Atif Rasheed said AI “can be used both positively and negatively” and called for stricter regulations to prevent its abuse. He called deepfake and explicit pornographic content “very disappointing” and argued for strict action against perpetrators. However, he rejected viewing the issue through a religious lens, saying the BJP “respects women of all religions” and the “Sulli Deals” and “Bulli Bai” cases were handled legally.

The Sulli Deals and Bulli Bai episodes occurred in 2021 and 2022, using doctored images. Both sparked widespread outrage and police investigations. Indian authorities arrested Aumkareshwar Thakur, alleged creator of the “Sulli Deals” account, and Niraj Bishnoi, identified as the creator of “Bulli Bai,” in January 2022. Both were released on bail on “humanitarian grounds” two months later. Researchers believe the rise of generative AI has significantly expanded the scale and speed of online harassment of Muslim women. New apps allow users to upload photos and automatically generate sexualized images. Such tools are widely available online, often free and requiring no technical expertise.

For those already living with persistent harassment, the emergence of AI-generated images has added a new dimension of fear. Afreen Fatima, 27, a researcher and activist who has faced online abuse since speaking out against India’s 2019 Citizenship Amendment Act (CAA), was among dozens of Muslim women whose photos were uploaded and “auctioned” on Sulli Deals. Four years after the Sulli Deals controversy, the abuse has barely let up. Anonymous accounts, often using common Hindu names, continue to reach out with abusive messages, rape threats and targeted harassment. “Every few days, there is a message from a random account with a rape threat or death threat,” she said. The prospect of AI-generated sexual imagery has added to that fear. Fatima said the online hate also shapes how she navigates public spaces. “I feel uncomfortable traveling alone,” she said. “When you see such fantasies circulating online about Muslim women, you start wondering if someone might attack you in real life.”

After the video went viral, Ayoub’s career opportunities began to disappear. “As a model, your reputation matters,” she said. “If negative comments appear on your profile, brands stop approaching you.” For four or five months, fake accounts flooded her profile with abusive comments, deterring potential clients. The harassment also changed her relationship with social media. “Instagram used to be my safe space,” she said. “Now I don’t feel safe there, and I limit what and how I post.” Ayoub reported the incident to the cybercrime unit of the police in New Delhi, filing a written complaint. “Nothing happened,” she said, adding that most abusive content was removed only because her friends reported those accounts en masse.

Legal experts say India’s current laws struggle to keep pace with AI-generated content. “The harm is real even if the image itself is fake,” said Apar Gupta, a lawyer and founding director of the Internet Freedom Foundation. Under Section 66E of India’s Information Technology Act, criminal penalties apply for capturing or publishing images of a person’s private areas without consent. But if the target’s body was never actually recorded – in other words, if the image is entirely AI-generated – this provision may not apply. “Even if the image is fake, it creates a permanent scarlet letter for women,” Gupta said. Digital platforms enjoy ‘safe harbor’ protections as long as they remove illegal material after being notified. But Gupta said many victims struggle even to achieve that. Without structural changes to platform design, algorithmic priorities and legal frameworks, AI abuse will continue to spread faster than any legal system can respond, he warned.

In such a scenario, accountability remains elusive for targeted Muslim women. “What I want most is to find the people behind those accounts,” Ayoub said. “They destroyed my reputation without even knowing me.”