Algorithmic microtargeting and AI deepfakes have significantly impacted voter behavior throughout elections within South Asian nations. Recent political campaigns are able to create highly personalized and emotionally charged content directed towards specific demographics. The proliferation of hyper-realistic cloned voices and manipulated videos makes it incredibly challenging for citizens to distinguish authentic news from falsehoods. Skepticism and targeted manipulation degrade public discourse, forcing voters to rely on bias rather than facts, eroding trust in democratic institutions.
Over the past few years, Artificial Intelligence has become more accessible and affordable to produce because its tools are now available to anyone. For instance, a 2024–2025 regional ITU survey highlighted that over 85% of South Asia’s urban areas now have high-speed mobile internet, giving millions of citizens instant, low-cost access to cloud-based AI generation platforms. AI-generated audio, videos, and images can make a person appear to be saying or doing something they actually never said or did. The danger of AI stems from its realism: familiar voices and faces can make viewers more likely to trust its contents before checking for authenticity. During elections, deepfakes can be used to spread false scandals, fabricate endorsements, imitate political leaders, or create misleading speeches.
Still, AI-generated content alone cannot explain the recent voting dynamics. Microtargeting is the practice of using voter data to send highly specific political messages to specific audiences. Campaigns may use information on an individual’s location, language, age, online interests, or community concerns to decide which advertisements or posts they view. Specifically, this can be seen in the difference between young voters, who receive messages about education or jobs, and rural voters, who may instead receive messages on local infrastructure or farming. Microtargeting can be seen as beneficial, as it could help campaigns to make sure that voters are reached with issues that matter to them. However, it becomes a harmful practice when campaigns use private data, make false claims, or send conflicting messages to separate groups without public accountability.
Artificial Intelligence and microtargeting impact voters’ perception of candidates, parties, and social groups. The first elicits an emotional reaction: creating fear, anger, or distrust; the second ensures that this emotionally triggering content is reached by individuals more likely to be responsive. This is suggestive of how political messages can be more emotionally persuasive than factual. Instead of viewing the same public campaign promises, different voters may receive completely different versions of a candidate’s message. An immediate consequence is the creation of echo chambers: an environment where most individuals view content that confirms and reinforces what they already believe in, without ever being challenged.
Throughout South Asia, elections take place in highly connected but politically divided digital environments. YouTube, TikTok, X, and WhatsApp allow political content to move more rapidly across languages and regions. Family group chats and community networks can make messages appear trustworthy because they are shared by people voters already know. At the same time, many users may not have access to reliable fact-checking resources in their preferred language. Bangladesh, Pakistan, and India all have large youth populations and active political communities, making digital outreach especially important. In this setting, AI can increase political participation, but it can also increase the speed and scale of misinformation.
India’s 2024 general election showed both the useful and risky sides of AI in campaigning. Political parties used AI to translate speeches into regional languages, create personalized videos, and revive the likenesses of deceased leaders for campaign messages. These uses can make campaigns more accessible to voters who speak different languages. However, Indian political parties also circulated manipulated audio, deceptive videos, and politically charged misinformation through social media and messaging apps. Because India has major religious and political divisions, false or targeted content can intensify already existing tensions. AI did not replace traditional campaigning such as rallies and door-to-door organizing, but it added another powerful way to influence voters and shape the information they receive.
In Bangladesh, digitally manipulated political content has become especially visible around election periods. Supporters of different political groups have circulated misleading posts, edited videos, and AI-generated material designed to make opponents look dishonest, unpopular, or dangerous. For instance, partisan networks used cheap AI avatar tools to circulate deepfakes of exiled opposition leader Tarique Rahman falsely admitting to corruption, alongside manipulated videos of female politicians designed to spark religious outrage. These messages often spread through Facebook and other social-media platforms, where political debate can already be intense. In the lead-up to Bangladesh’s 2024 elections, the growth of AI-made political videos made it more difficult for voters to know whether clips were real, satirical, or intentionally deceptive. When voters cannot tell the difference, they may become cynical about all political information, including legitimate reporting. This can reduce informed participation and make it easier for false narratives to influence public discussion.
Pakistan’s 2024 election provided a notable example of AI being used as a campaign tool rather than only as disinformation. While former Prime Minister Imran Khan was imprisoned, his party, Pakistan Tehreek-e-Insaf (PTI), used AI to create audio and video messages based on words he had communicated from jail. The party openly identified the recordings as AI-generated and used them to reach supporters through online rallies and social media. This showed how technology can help a restricted political movement communicate with voters. However, the same tools can also be used deceptively to imitate political figures or spread false claims. In Pakistan’s already polarized political environment, voters must decide whether online content is authentic, officially authorized, or fabricated by an opponent.
The wider effect of deepfakes and microtargeting is a decline in shared trust. Democracies depend on citizens having access to enough reliable information to debate public issues and make informed choices. When voters receive separate, emotionally charged messages and cannot tell whether video or audio is real, public discussion becomes more fragmented. People may distrust journalists, election officials, and even authentic recordings. This phenomenon is sometimes called the “liar’s dividend”: once deepfakes are common, real evidence can be dismissed as fake. The result is not simply misinformation; it is a society in which people may no longer agree on basic facts.
AI deepfakes and microtargeting are changing elections in India, Pakistan, and Bangladesh by making political communication faster, more personalized, and more difficult to verify. These technologies can have positive uses, such as translating campaign messages, increasing access to leaders, and helping voters learn about local candidates. Yet they can also spread falsehoods, exploit divisions, and weaken trust in reliable information. Protecting democratic elections will require clearer platform rules, transparent political advertising, independent fact-checking, digital-literacy education, and responsible use of AI by political parties. Ultimately, technology should help citizens make informed choices rather than manipulate their fears or deepen social divisions.
Anwitha Kandula
- Anwitha Kandula
