The recent Bihar state elections showed how artificial intelligence has become a major tool in political campaigning. AI was used to make speeches, videos, and messages that targeted voters quickly and at a low cost. Voice cloning in local dialects, chatbot conversations, and deepfake videos became common across social media, WhatsApp, and Telegram.
While this helped candidates reach people in remote areas, it also created confusion among voters because it was difficult to understand what was real and what was AI-generated. Fact-checkers and officials found it tough to control the spread.
Artificial Intelligence Became The Core Of Campaigning
According to a report from Rest of World, political parties in Bihar depended heavily on artificial intelligence for almost every part of their election strategy.
Manish Kumar Prasad, who runs Lemon Dot Media, said his team used tools like ElevenLabs’ voice generator, ChatGPT, and Claude to create speeches in local dialects and short videos for different groups of voters.
These messages travelled fast through WhatsApp and Telegram groups and helped campaigns reach nearly 75 million voters.
Prasad shared that voice cloning was the most demanded service because it made voters feel like the candidates were speaking directly to them. This strategy reduced travel, manpower, and cost.
Many candidates saved money by paying for AI subscriptions instead of hiring large communication teams. Chatbots also answered voter questions in their local dialects and continued even during the mandatory 48-hour silence period when most campaigning activities are banned.
Deepfakes & Misinformation Became A Serious Risk
Deepfakes and AI-generated audio created major confusion among voters. Videos showed politicians campaigning in places where they were not physically present, and fake promises and endorsements circulated widely. Younger voters could sometimes spot AI content, but many older citizens believed it to be real.
Police formed a 15-member team trained in AI to monitor content day and night. Nearly a dozen cases were registered for illegal material, but many clips went undetected because they looked real.
Fact-checkers said that at least one in five election-related posts was AI-generated. Critics argued that AI gives an advantage to wealthier parties who can afford advanced tools, leaving smaller candidates behind.
