The Role of Generative AI in Intensifying Social Marginalization Via Disinformation

• In Event: AI, Equity, Ethics, and Inclusion

Abstract

While disinformation has long served as a propaganda tool, technological advancements—including the internet, social media, and artificial intelligence—have dramatically widened its scope and accelerated its spread. Generative AI (GenAI) in particular threatens to severely pollute the information environment by enabling the rapid creation, modification, and curation of false narratives. When directed at vulnerable social groups, such narratives can trigger hateful rhetoric, social mobilization, and direct violence, deepening the marginalization of already vulnerable communities. This paper examines how the unchecked use of GenAI facilitates the creation and reinforcement of harmful narratives in the Indian context, focusing on the intersection of GenAI, disinformation, and social vulnerability. In India, religious minorities, women, Dalits, and other marginalized groups are frequently targeted by disinformation campaigns.

We analyze over 3,000 claims from Indian fact-checking organizations—including The Quint, AltNews, FactChecker, and BoomLive—spanning 2012 to 2026. The data reveals an exponential rise in disinformation campaigns between 2022 and 2025, closely tracking the release of ChatGPT 3.5 in November 2022. Since then, GenAI-assisted disinformation has grown sharply; by 2025, nearly 61% of all disinformation campaigns used GenAI in some form, and currently approximately 30% of all claims in the dataset involve GenAI tools. Of all claims analyzed, 81% are false, 7.3% misleading, 3.7% partially true, and only 8.4% true. The most frequently identified beneficiary of disinformation is BJP/NDA, India's ruling political coalition. Approximately 26% of claims target religious minorities (Muslims, Christians, or Sikhs), while Dalits, Tribals, women, and the elderly are also frequent targets. In many cases, disinformation campaigns are followed by hate speech, calls for violence, and physical mobilization against these groups. This is corroborated by data from the Centre for Study of Society and Secularism (CSSS), which recorded 59 communal riots in India in 2024, up from 32 in 2023—a trend that correlates with the rise in GenAI-aided disinformation. Using a mixed-methods approach, this research qualitatively codes the nature of disinformation and quantitatively analyzes its relationship with GenAI use, communal violence, and mob lynchings. Scatter plots, cross-tabulation, and crisp-set QCA indicate positive correlations between the rise of GenAI, disinformation campaigns, communal riots, and mob lynchings. The evidence suggests that GenAI is being actively used to vilify marginalized communities—portraying them as undesirable, suspicious, or threatening—thereby intensifying their structural vulnerability.

The paper concludes with a call for further research and robust policy reforms, emphasizing the need for transparent and accountable media, restraint among public officials, and responsible citizenship.

Author

• Shahidur Rashid Talukdar, Virginia Polytechnic Institute & State University
  Presenting Author