AI risks distorting information but also holds untapped potential to improve electoral campaigns and processes. To this end, prebunking must be used by all stakeholders
Bihar’s political landscape is gradually heating up in the wake of the upcoming Assembly polls with political parties crafting campaigns to woo voters. The campaigns will soon be dominated by artificial intelligence (AI)-generated videos and audios.
In March, a status report submitted by the Ministry of Electronics and Information Technology (MeitY) to the Delhi High Court raised concerns on the harmful use of AI during elections, and pressed for mandatory regulation around AI content disclosure. This warrants a deeper look at the role of AI during polls, and how its risks can be mitigated.
Unprecedented electoral activity across the world in 2024 coincided with the rapid advancement of AI, sparking widespread debate about its potential to reshape political campaigns, voter engagement, and electoral management. While AI offered innovative tools to enhance efficiency and outreach, it also raised concerns about misinformation, deepfakes, and ethical challenges. India also witnessed the use of AI in election campaigns for voter profiling, targeted messaging, misinformation/disinformation, and generating deepfakes. Many feared that AI when used in elections would result in catastrophic outcomes, but only 2% of the content was AI generated, largely used to augment outreach and efficiency of voter engagement.
AI was used to generate fake audios/videos of celebrity endorsements, satirical narratives, and propaganda images/videos. Moreover, dead politicians were ressurected through deepfake videos to invoke nostalgia and loyalty among voters, which raised ethical concerns . Such use of generative AI was without ample oversight and transparency.
The Election Commission of India (ECI)’s response was too little, too late. The ECI’s AI guidelines issued mid-way through the polls lacked clarity and deterrence. Engagement with stakeholders such as independent fact-checkers, and journalists reduced as the polls progressed.
Even though the impact was not as concerning as predicted, these elections acted as a testing ground for future amplification of AI. This poses significant risks as the information ecosystem continues to remain distorted. The 2024 Global Risks Report considers India a high-risk country for misinformation and disinformation.
A Social Media Matters study found that 80% of first-time voters came across fake news, with WhatsApp accounting for the largest share (30%). During the 2025 Delhi polls, a survey by The 23 Watts revealed that 91% of respondents under 25 years believed fake news could influence voting, 80% admitted it shaped their perceptions, 59% were influenced by sensational content, and 45% shared unverified ‘news’. While the immediate impact may seem limited, long-term effects could affect election results and erode trust in institutions.
Thus, we need to adopt strategies that can reduce AI-driven exploitation, and identify how generative AI can strengthen the polls. Political parties must adopt ethical guidelines for use of AI in campaigns. Parties and tech platforms need to follow an ECI mandated code of conduct regulating AI-generated content used for campaigns.
AI-driven chatbots and recommendation systems personalise voter outreach, significantly cutting campaign costs. AI-powered robocalls — eight times cheaper than human call centres — were widely used in the 2024 polls, cloning local politicians’ voices for a more personal touch. Over two months prior to the general elections, more than 50 million AI-generated calls were made. For ethical transparency, these calls must have included disclaimers stating they were AI-generated.
Beyond campaigns, AI is transforming electoral management, helping the ECI streamline processes and enhance transparency. It uses AI to maintain and update India’s massive electoral roll. Machine learning algorithms detect duplicate entries, deceased voters, or inconsistencies in voter IDs. This ensures a cleaner, more accurate voter list, thus reducing chances for fraud.
Misinformation detection is another vital use case of AI. Real-time monitoring systems scan social media for fake news, or doctored videos, flagging them for review. The ECI collaborates with tech firms to deploy natural language processing models that identify misleading narratives, ensuring a fairer information ecosystem. AI can optimise distribution of EVMs and polling resources with predictive models forecasting EVM demand across constituencies. AI-driven voter education is another key area to combat misinformation, and boost awareness.
AI risks distorting information but also holds untapped potential to improve electoral campaigns and processes. To this end, prebunking must be used by all stakeholders as it helps individuals develop a more sceptical, critical mindset, and build resilience to misinformation.
Finally, the onus lies on political parties for employing generative AI for ethical purposes that can enhance their voter outreach capabilities at reduced costs without manipulating their emotional responses. The ECI must remain proactive in its response to unethical use of AI. It must issue timely directives that have legal implications and build resilience against electoral misinformation by amplifying trusted voices.
The article was originally published in Deccan Herald.