In November last year, Muralikrishnan Chinnadurai noticed something unusual while watching a Tamil-language event livestreamed from the UK. A woman introduced as Duwaraka, daughter of the late Tamil Tiger militant leader Velupillai Prabhakaran, was giving a speech. This was odd because Duwaraka had reportedly died in a 2009 airstrike during the Sri Lankan civil war, and her body was never found. Chinnadurai, a fact-checker in Tamil Nadu, quickly identified glitches in the video, revealing it as an AI-generated figure. He was concerned about the potential for this misinformation to spread, especially with upcoming elections.
As India approaches its elections, AI-generated content is becoming ubiquitous, with campaign videos, personalized audio messages, and automated voter calls proliferating. Content creators, like Shahid Sheikh, are using AI tools to show politicians in unusual scenarios, such as wearing casual attire or dancing. While this can be entertaining, experts worry about the implications of making fake news appear real. SY Qureshi, India's former chief election commissioner, warned that rumors spread rapidly on social media could have severe consequences.
AI's use in politics is not unique to India; for instance, jailed Pakistani politician Imran Khan addressed a rally using AI. In India, Prime Minister Narendra Modi has effectively used AI to campaign, employing a government-created AI tool, Bhashini, to translate his speeches into Tamil in real time. However, AI can also manipulate messages. Recently, Bollywood stars Ranveer Singh and Aamir Khan filed police complaints about deepfake videos campaigning for the opposition Congress party, made without their consent. Prime Minister Modi himself raised concerns about AI-distorted speeches from ruling party leaders. This led to the arrest of two individuals from opposition parties over a doctored video of Home Minister Amit Shah.
Despite these incidents, there is no comprehensive regulation for AI use in India. According to data and security researcher Srinivas Kodali, the lack of regulation means violators might only face minor repercussions. Creators like Divyendra Singh Jadoun, founder of The Indian Deepfaker (TID), stress the importance of ethical practices, such as disclaimers on AI-generated content. Yet, the ease of creating deepfakes complicates control efforts. Marketing agency worker Mr. Sheikh has seen his AI-created images used without permission by politicians.
The ease of creating deepfakes—now achievable in minutes—poses significant risks. The Indian government initially refrained from regulating AI but acted in March after controversy over Google's Gemini chatbot response labeling Modi a fascist. The government now requires tech companies to seek permission before launching "unreliable" generative AI models and warns against responses that threaten electoral integrity. However, fact-checkers like Chinnadurai find it challenging to keep up with misinformation during election periods. Kodali notes that AI-generated fakes are infiltrating mainstream media, while the election commission remains silent on AI regulation.
Experts believe there is no foolproof solution yet. Qureshi suggests that penalizing those who forward fake content might deter the spread of unverified information.