Opinion
The proliferation of deepfakes has shrouded India’s 2024 polls in uncertainty
Summary
- The online spread of deceptive AI-generated content aimed at voters could hurt Indian democracy now that the menace of fake clips has grown so real.
Almost every Indian election in the past few decades has integrated new technology into the campaign process, from phone calls in the 1990s to holograms in 2014. Poll season in 2014 was dominated by Facebook and Twitter. In 2019, the election campaign centred around WhatsApp. It’s evident that voters can be tricked and perplexed by ever-evolving technology. An increase in disinformation that social media platforms can disseminate can have disturbing consequences. Prime Minister Narendra Modi referred to deepfakes as a “new age sankat" or difficulty in a recent speech. How might deepfakes affect important elections and the trajectory of geopolitics? Is artificial intelligence (AI) equipped for it?
Slovakia’s parliamentary elections in September marked a turning point in the history of elections affected by AI. Two days prior to polling, an audio recording was posted on Facebook in which a journalist and the liberal Progressive Slovakia party’s leader seemed to discuss ways to rig the election. Both denounced the audio as fake right away and fact-checking revealed evidence of AI manipulation. But once the votes were counted, Progressive Slovakia had fallen short in a close contest.
Although political propaganda and the dissemination of misinformation have been around for centuries, AI-generated videos, audio clips and images now have a greater potential to impact electoral choices because it has become very hard to tell what is real apart from what is fake. The 2023 Slovak election is widely dubbed a “test case" in the context of a series of major elections due in 2024, including those in the US, UK, EU and India (Lok Sabha polls).
AI was also extensively used in campaigns by candidates in Argentina’s recent presidential elections. In a New York Times article, Jack Nicas and Lucía Cholakian wrote that “with its expanding power and falling cost, [AI] is now likely to be a factor in many democratic elections around the globe."
Deepfakes are not a new-entrant in the Indian scenario. During the 2020 Delhi assembly elections, videos of then state Bharatiya Janata Party (BJP) president Manoj Tiwari criticizing chief minister Arvind Kejriwal’s policies in different languages were widely circulated, raising an alarm over AI meddling in elections. Massachusetts Institute of Technology (MIT) ascertained that the videos were produced by AI.
Three years is a considerable time for such technologies. Recently, in election-bound Madhya Pradesh and Telangana, a few doctored video clips went viral. One such video—which featured questions from the ‘Kaun Banega Crorepati’ (KBC) quiz show and was crafted to portray the BJP’s Madhya Pradesh leader Shivraj Singh Chouhan in a negative light—demonstrated how deepfakes can be used as a weapon in political warfare. The problem will only worsen, given the technology’s rapid improvement. Packages of customized political advertising campaigns can reportedly range from $199 to $599!
Enthusiasts may even spread readily available and inexpensive political deepfakes without the campaign teams of parties being aware of them. Recall how a recent image of Donald Trump’s stormy arrest, produced by a generative AI tool called Midjourney, went viral. How deceptive can such deepfakes be during an important election? In her 2020 book Deepfakes: The Coming Infocalypse, author and AI expert Nina Schick observed that—with advancements in AI, video creation and online trolling—deepfakes not only pose a real threat to democracy, but can also escalate voter manipulation to unprecedented heights.
Before the Lok Sabha elections next year, the quality of deepfakes will undoubtedly improve significantly. According to The State of Deepfakes report brought out by a cybersecurity firm, India is the sixth most vulnerable country to deepfakes. Apprehensive of AI-generated deepfakes and misinformation, the Indian government has announced plans to take action.
However, as technology advances, it might become impossible to identify all deepfakes in real time. Furthermore, any deceptive deepfake posted shortly before an election—like the one in Slovakia—might reach millions of people and could have already done enough harm by the time it’s proven to be fake.
However, it’s still uncertain how deepfakes could affect an election. Do the voters really cast their votes based on an audio or video clip? Or are those most influenced by deepfakes predisposed to being biased in a certain way? Deepfakes are cheap and simple to create and can be made by anyone to post on social media. In Argentina, the campaigns of both main contenders are said to have heavily disseminated deepfakes. I couldn’t find any reliable statistical analysis or survey to determine the extent of potential harm caused by deepfakes. Is it possible to determine the precise quantitative impact of the fake KBC clip on the outcome of the recent Madhya Pradesh assembly elections that the BJP won? It’s clearly not simple. To determine the extent to which we should be concerned, however, a proper data-based assessment is required. What can be said at this point is that deepfakes have added a shade of uncertainty as we inch closer to the 2024 Lok Sabha elections.
