Marketing Research
What India’s Elections Can Teach Us About AI
Executive Summary
India just wrapped up the world’s largest elections during which we saw broad use of AI, offering insights for future election cycles and AI’s possible role in them. The elections, which closed earlier this month, gave Prime Minister Narendra Modi another term but denied his Bharatiya Janata Party (BJP) party a parliamentary majority.
During the election cycle, political parties in India used AI in several ways, mostly without repercussions, and with little transparency. AI helped candidates and political parties scale their messaging to a diverse, multilingual electorate but also spread misleading information.
- AI helped parties generate satirical images and videos targeting political opponents, create content that boosted politicians’ images, resurrect popular politicians and figures to support one party over another, and spread false narratives about rivals.
- Parties also leveraged AI to broaden the scope of their outreach to voters, especially those in far-flung areas or where language differences have historically posed access challenges.
The multifaceted use of AI in India’s elections—ranging from seemingly simple to sophisticated use cases—offers valuable insights as we prepare for upcoming elections around the world. Individuals are using AI to create satirical images, videos, and memes, which can serve a dual purpose of disseminating influence messaging, even mis/disinformation, under the guise of humor. Some candidates are embracing the use of AI-generated digital avatars and voice cloning to enhance their messaging efforts. Political campaigns are seeking ways to use AI to improve their voter outreach and create political advertisements. Collaboration between AI developers and political parties or candidates will possibly expand—Indian AI companies that led the charge on AI-generated election content are now seeking to expand their business globally, including to Canada and the United States. Meanwhile, adversaries are using free and subscription AI tools to generate content and manufacture inauthentic engagement on social media.
As new AI capabilities emerge, partnerships and joint efforts are required to stay abreast of developing tactics and use cases. Nisos works with a variety of partners to combat harmful usage of AI within the context of elections, influence operations (IO), and from platform security perspectives, helping our stakeholders identify AI-generated and -altered content, develop technical mitigations, and properly contextualize and scope the impact of AI to the information environment.
Overview
India’s six-week election cycle closed on 1 June and saw 642 million voters cast ballots. India has a history of using digital content in its political campaigns. In 2014, for example, Prime Minister Modi created a 3D hologram version of himself to appear at numerous campaign rallies (see source 1 in appendix). In 2022, a political consultant created a doctored video for a BJP politician in Delhi by using lip-sync technology to encourage people to vote for the BJP. This year’s election occurred against the backdrop of remarkable growth of India’s AI market and heightened concerns around AI-generated content.
Nisos has been monitoring this year’s Indian elections as a case study to analyze how candidates and political campaigns are using AI to disseminate messaging to a diverse electorate in multiple languages at scale. As democracies around the world prepare for how AI may be used to enable election-related IO, we also consider the productive potential that AI technologies offered to Indian voters this year. We identified AI-generated and AI-manipulated content presenting misleading information and false narratives, as well as examples where AI supported enhanced voter access and outreach efforts.
Use of AI During The Indian Election
Election Candidates Used AI to Create Digital Content
Across the political spectrum in India, we saw political parties use AI to create video and audio deepfakes (see source 2 in appendix) as satire about other candidates, to enhance their own image, and to revive deceased public figures to speak to voters. Few repercussions, either from official action or the public, resulted from the dissemination of such content. Limited transparency around the use of AI-generated content created confusion among voters in some cases. Indian political parties also collaborated directly with AI developers and startups, spending an estimated $50 million on AI-generated campaign material (see source 3 in appendix). Examples of AI-generated content include:
- Satirical Content Targeting Opponents: The BJP circulated an AI-generated video of one of Modi’s leading opponents, who was jailed at the time, strumming a guitar and singing a verse from a popular Bollywood song. (see source 4 in appendix) Another video showed a prominent Indian Muslim politician singing devotional Hindu songs. The opposition India National Congress party also posted a video parodying Modi by rewriting song lyrics to describe how Modi was handing over the country to business interests. (see source 5 in appendix)
- Enhancing a Candidate’s Image: Indian influencers on social media with large followings generated their own content using AI, which candidates in some cases reposted. For instance, Modi reposted an AI-generated video clip of him dancing with the hashtag PollHumour (Graphic 1) (see source 6 in appendix). The video generated over 18.6 million views on X (formerly Twitter). An AI-generated avatar of Modi that individuals shared on messaging apps shows him addressing a series of voters by name. (see source 7 in appendix)
- False Narratives: BJP supporters posted videos with false narratives and misinformation about rival political party members. One account shared a video of Indian National Congress leader Rahul Gandhi with AI-generated audio announcing Gandhi’s resignation from the party (Graphics 2 and 3). Despite significant media attention and citations of this event in coverage of election-related AI, the original posts containing the generative AI audio only received approximately 9,000 and 3,000 views. (see source 8 and 9 in appendix) Another video from late April using AI voice-cloning of Gandhi shows him being sworn in as prime minister. (see source 10 in appendix)
- Digitally Resurrecting Popular Figures: The BJP used AI to recreate the voice of a famous singer overlaid on videos of Modi campaigning. (see source 11 in appendix) The Dravida Munnetra Kazhagam party in Tamil Nadu worked with an AI firm to create a video using AI to revive a deceased party leader and have him endorse the current state government (Graphic 4).(see source 12 in appendix) Another AI-generated audio clip featured a former politician who died in 2016 criticizing the current state government. (see source 13 in appendix)
Graphic 1: Screenshot of an AI-generated video Prime Minister Modi
posted showing him dancing on stage at a concert. (see source 14 in appendix)
Graphics 2 and 3: X posts featuring generative AI audio of Rahul Gandhi announcing his resignation. The posts only garnered approximately 12,000 combined views. (see source 15 and 16 in appendix)
About Nisos®
Nisos is the Managed Intelligence Company. We are a trusted digital investigations partner, specializing in unmasking threats to protect people, organizations, and their digital ecosystems in the commercial and public sectors. Our open source intelligence services help security, intelligence, legal, and trust and safety teams make critical decisions, impose real world consequences, and increase adversary costs. For more information, visit: https://www.nisos.com.