India Elections: Bollywood Stars Targeted in Deepfakes
Facts
- In India, deepfake videos in which two Bollywood stars are seen criticizing Prime Minister Narendra Modi and asking people to vote for the opposition have gone viral online since last week, prompting worries about the role artificial intelligence (AI) can play in the elections.1
- On Monday, Indian actor Ranveer Singh filed a complaint and First Information Report with the Mumbai Police's Cyber Crime Cell as a 41-second video of him at a fashion show in Varanasi was edited with fake audio.2
- This comes after Aamir Khan reported tampered videos to relevant authorities and also filed a police complaint, as clips from the show he hosted between 2012 and 2014 were doctored using AI.3
- India's multiphase general elections—the most expensive in the world—began last Friday and will run until June 1, allowing nearly 1B people to cast votes for the 543-seat parliament.4
- Electoral deepfakes have gained attention worldwide recently, with reported incidents ahead of national elections in Bangladesh, Nigeria, and Slovakia, as well as in the New Hampshire Democratic presidential primary.5
- In India, however, the first known political use of a deepfake video dates back to the 2020 Delhi Legislative Assembly elections, when the ruling Bharatiya Janata Party (BJP) dubbed videos in different languages to connect with specific demographic voters.6
Sources: 1Reuters, 2The Hindu, 3Independent, 4New York Times, 5The Times of India and 6The Diplomat.
Narratives
- Pro-establishment narrative, as provided by NPR Online News. Online disinformation has long been a factor influencing the outcome of elections around the world. However, deepfakes may prove to be a game changer for democracies — particularly at the local level, as the collapse of local journalism makes it harder to check information and debunk false content.
- Establishment-critical narrative, as provided by Dazed. While AI has facilitated the spread of disinformation, the effect of such campaigns on influencing votes is less than typically assumed—especially in highly polarized environments. While AI can contribute to distrust in election results, overreacting with panic and alarmism about this technology can produce the same or an even worse outcome.