Simulated Inclusion: AI and the politics of visibility in India’s democracy

The above example is only one among many that have emerged since AI became an integral part of political campaigning following India’s general election in 2024. While AI-generated electoral content is often difficult to detect and poses broader risks to democratic institutions, it also creates particular vulnerabilities for marginalized groups. One such risk is what I describe as “simulated inclusion.” This refers to a form of inclusion in which individuals, especially women already facing intersectional barriers related to caste, religion, gender, or poverty, appear to gain political participation through AI-driven targeted outreach, without any corresponding increase in agency, accountability, substantive voice, representation, or legal protection.

This dynamic can be observed, in the use of AI-generated voice-cloning tools to translate political speeches into local dialects. What constitutes misinformation may initially appear as an inclusive practice, offering marginalized groups greater access to political information and making them feel recognized or “seen.” Yet such practices merely simulate inclusion by relying on identity-based micro-targeting and by exploiting existing inequalities in digital literacy, access to resources, and the ability to distinguish between authentic and AI-generated content. Lower-income and lower-caste communities, for instance, often rely heavily on mobile-based messaging platforms as their primary source of news.

Simulated inclusion also operates through the mobilization of specific identities and social positions. The earlier example of the AI-generated video depicting Prime Minister Narendra Modi and his late mother relied on grief, intimacy, and moral appeal to shape political perception. While the video circulated broadly, its emotional registers resonated particularly strongly with economically and socially marginalized Hindu women, for whom familial devotion, motherhood, and sacrifice carry deep cultural significance. Such targeting does not empower these women politically; it exploits the very values that define their social position, producing visibility without voice: the essence of simulated inclusion.

Messaging tailored to local languages, cultural idioms, and community narratives offers recognition and visibility, yet this visibility does not translate into empowerment. Instead, it increases exposure to manipulation, reputational harm, and political polarization. As such, simulated inclusion captures a core paradox of AI-mediated democracy: those who are already structurally disadvantaged become the most intensively targeted and are drawn into political communication networks under deeply unequal conditions, yet the same people remain the least protected. These dynamics are particularly potent in an environment marked by uneven digital literacy and persistent caste- and gender-based inequalities which arise in the lack of evident ability to evaluate political information.

With approximately 80 per cent of first-time voters exposed to mis- or disinformation widely sharing false content unknowingly, the World Economic Forum’s Global Risks Report 2024 unsurprisingly identified India as the country most vulnerable to disinformation. Rather than correcting democratic deficits, AI-mediated political campaigning risks deepening these deficits by reproducing existing hierarchies under the guise of participation and technological progress.

India’s response to the harms caused by the Congress “case” has been insufficient. Although the takedown order issued by the Patna High Court marked a step forward, it came only after the damage had already occurred. The court was compelled to involve global platforms such as X and Google, underscoring the inability of existing legal frameworks to keep pace with the speed, scale, and transnational reach of AI-generated political content.

More broadly, courts in India and elsewhere face significant challenges in addressing AI-driven disinformation and deepfakes, particularly when it comes to attributing intent and responsibility for manipulated content circulating across encrypted and platform-mediated environments. Accountability is fragmented among content creators, digital intermediaries, and political actors, leaving significant regulatory gaps.

While courts rely on Article 21 of the Indian Constitution (right to privacy and human dignity), the Representation of the People Act, the Information Technology Act, and the Election Commission guidelines to address misinformation, these frameworks remain largely reactive. They focus on unlawful content and identifiable speakers rather than on the systemic risks posed by algorithmic amplification, profiling, and targeted political messaging. As a result, technological power remains largely unchecked, enabling political actors to exploit platform infrastructures to disseminate misleading content to marginalized communities with minimal scrutiny. This allows exclusionary narratives, gendered asymmetries, and forms of simulated inclusion to proliferate while remaining formally lawful.

Addressing these challenges requires a fundamental rethinking of digital governance. Comparative regulatory approaches offer useful guidance. The EU’s Digital Services Act and Artificial Intelligence Act shift the focus from individual pieces of content to systemic risks by requiring platforms to assess and mitigate harms arising from recommender systems, political advertising, and profiling practices.

In practice, this entails obligations to evaluate how algorithmic systems may amplify misinformation, discrimination, or political manipulation, alongside enhanced transparency regarding content prioritization, ad targeting, and user redress mechanisms. In December 2023, the European Commission opened formal proceedings against X for potential breaches related to risk management, transparency, and the design of platform systems that may facilitate the spread of illegal or misleading content.

These developments are not a blueprint for India but illustrate a different regulatory logic that targets systemic platform risks rather than isolated pieces of content. In the Indian context, where electoral politics operate amid deep social hierarchies, linguistic diversity, and uneven digital literacy, governance responses must move beyond reactive takedowns and instead develop context-sensitive mechanisms for transparency in political advertising and oversight of algorithmic campaigning.

Recent regulatory developments in India illustrate this tension sharply. The government’s new rules requiring social media platforms to comply with takedown orders within three hours, and to permanently label AI-manipulated media, are framed as necessary responses to deepfakes and digitally mediated abuse. While the rise of synthetic sexualized imagery and political disinformation undoubtedly poses serious risks, especially for women from marginalized communities, the compressed timeline for removal raises concerns about automated censorship and the erosion of digital freedoms. In contexts already marked by caste and religious polarization, rapid takedown regimes may simultaneously expand state control over speech while failing to address the structural hierarchies that make certain women more vulnerable to digital harm in the first place.

Any future regulation of AI in elections must be grounded in India’s constitutional commitments to equality and social justice, including guarantees of equality before the law and non-discrimination under Articles 14 and 15, as well as protections for linguistic diversity. AI governance should therefore move beyond abstract concerns with electoral integrity and address how caste, religion, and gender intersect to shape unequal exposure to political harm. This requires an agency-centred approach that prioritizes substantive political voices over mere visibility, supported by transparency in algorithmic targeting and accessible legal and political remedies in local languages. Without such grounding, AI regulation risks reproducing the very exclusions it seeks to address.

Ultimately, digital democracy must offer genuine empowerment rather than a technologically mediated performance of participation that extracts data while perpetuating inequality. When AI only creates the illusion of inclusion, it does not strengthen democratic processes; instead, it exposes already vulnerable populations to new and intensified forms of structural harm. India’s experience despite its status as the world’s largest democracy serves as a critical warning.

In societies marked by deep social heterogeneity and persistent hierarchies, the deployment of advanced AI in electoral contexts risks intensifying, rather than resolving, democratic deficits.

As AI increasingly reshapes political campaigning worldwide, India stands as a crucial stress test for how democratic participation can be undermined under conditions of entrenched inequality, with implications that extend far beyond its borders.