TLDR: A new report by the Center for the Study of Organized Hate (CSOH) indicates a significant surge in AI-generated images and videos being used to spread anti-Muslim propaganda in India. The study, covering over 1,300 posts from May 2023 to May 2025, highlights four dominant hateful narratives, with sexualized depictions of Muslim women being the most engaged content. The report warns of severe implications for India’s Muslim minority and criticizes social media platforms for failing to remove flagged content.
NEW DELHI, India – A recent comprehensive report by the Washington-based Center for the Study of Organized Hate (CSOH) has unveiled a disturbing trend: the systematic deployment of Artificial Intelligence (AI) text-to-image tools to generate and disseminate anti-Muslim hate content across India. The think tank warns that generative AI is acting as a “force multiplier” for Islamophobic propaganda, posing significant risks to India’s 200 million Muslims and threatening the nation’s secular foundation.
The study, titled “AI-generated Imagery and the New Frontier of Islamophobia in India,” analyzed 1,326 publicly available AI-generated images and videos. These were collected from 297 public accounts posting in Hindi and English across major social media platforms including X (formerly Twitter), Instagram, and Facebook, between May 2023 and May 2025. The report found that the bulk of this hateful content circulated after mid-2024, coinciding with the increasing popularity and accessibility of AI tools such as Midjourney, DALL·E, and Stable Diffusion. Collectively, these posts garnered more than 27 million engagements.
Researchers identified four dominant narratives within the AI-generated content. The most prevalent and highly engaged theme involved the sexualized depiction of Muslim women, attracting 6.7 million interactions. This highlights a gendered aspect of Islamophobic propaganda, merging misogyny with anti-Muslim sentiment by portraying women as targets for harassment or violence.
Conspiratorial narratives formed another major category, with AI-generated visuals amplifying false claims like “Love Jihad” (the assertion that Muslim men seduce Hindu women for conversion), “Population Jihad,” and fabricated stories such as “Rail Jihad.” Such content aims to portray Muslims as a demographic and security threat to Hindu society. Additionally, exclusionary and dehumanizing rhetoric, often employing animal imagery to depict Muslims and advocating for violence or exclusion, constituted nearly a quarter (23.4%) of the analyzed posts. The aestheticization of violence was also noted, where AI was used to normalize and legitimize anti-Muslim aggression.
Nabiya Khan, a co-researcher of the report, emphasized the alarming nature of this development, stating that these AI tools have made “old prejudices scalable, faster, cheaper, and harder to trace.” The report underscores that this proliferation of hateful AI-generated content “threatens to further colonize the Indian information sphere, which is already marked by rampant misinformation, anti-minority bias, and a severe crisis of credibility.”
The study also pointed out the critical role played by Hindu nationalist media outlets, such as OpIndia, Sudarshan News, and Panchjanya, in embedding synthetic hate into mainstream discourse. This amplification ensured the content reached audiences far beyond extremist online communities.
Also Read:
- AI Chatbot Grok’s ‘White Genocide’ Responses Raise Concerns Over Generative AI Weaponization
- Generative AI Propels Sophisticated Academic Fraud in Chinese Paper Mills
Despite the clear violations, social media platforms have largely failed to take action. Out of 187 flagged posts, none were removed, indicating a systemic failure by these platforms to enforce their own community guidelines. In response, CSOH has urged a multi-pronged approach from lawmakers, platforms, and civil society, recommending updates to India’s Information Technology Act and intermediary rules to specifically address AI-generated images and videos, and mandating provenance metadata in all AI outputs to trace synthetic content back to its source.
