Fake alert: Voters need to be alert against deepfakes in election campaigns
Though manipulated information has often been tabled by parties in the past, voters are finding it challenging to decipher deepfakes as real or fake.
Fact and fiction have become synonyms as AI tools manufacture information quicker than the speed of thought. At the crucial juncture of electing its next leader, the Indian voter now has, the additional, task of distinguishing between the truth and a synthesised version of it.
Rapid technological advancements have enabled political parties to freely employ deepfakes while campaigning during the ongoing elections. Though manipulated information and deceit have often been tabled by parties in the past, deepfakes are hyper-realistic audio, images, or videos generated using sophisticated algorithms challenging the human capabilities to decipher them as real or fake.
Deepfake Misinformation Blitzkrieg
Videos with unclear sources have surfaced depicting Prime Minister Narendra Modi addressing a series of Indian voters by name. The precision targeting enabled by AI allows campaigners to tailor avatar videos or audio calls in any of India's myriad regional languages and dialects to connect more directly with diverse voter blocs.
On May 6, Modi's X account reshared an AI-generated video of him rocking a stage on a Bollywood track, stating, "Such creativity in peak poll season is truly a delight."' Social media is drenched in videos of Modi tuning to various Bollywood songs and trends — conceivably boosting his popularity amongst young voters.
The use of AI is not limited to impersonation of politicians. In viral videos, Bollywood actors Aamir Khan and Ranveer Singh can be seen delivering a critique of Modi's tenure. The AI-synthesised doppelgangers accuse the incumbent of failing to deliver on key campaign promises and inadequately addressing pressing economic woes over his two terms at the helm. Both the actors have filed police complaints against the viral clips.
The 2024 Global Risks Report by the World Economic Forum suggest that Indian experts flagged misinformation and disinformation as the "biggest threat" to their country in the next two years. The generative AI ordeal only exacerbates such threats.
Tech-Fuelled Hate Narratives
Another tech-enabled campaigning move that raised concerns was a provocative animated video put out by BJP Karnataka's social handle. The clip deployed caricatures to depict Congress leader Rahul Gandhi and Karnataka chief minister Siddaramaiah placing an egg labelled "Muslims" alongside three others marked "SC" (Scheduled Castes), "ST" (Scheduled Tribes), and "OBC" (Other Backward Classes) in a nest.
The disturbing allegory showed Gandhi selectively "feeding funds" to the egg representing Muslims, which then hatched into a bird donning a skullcap. As the Muslim bird emerged, it shoved away the nascent fledglings from the SC, ST, and OBC eggs while ominous laughter played in the background — crudely implying partisan favouritism towards the Muslim community at the expense of other minority groups.
Though targeted at the Congress party, the arrow seemed to have landed on the Muslim community equally.
Nitasha Kaul, a professor of politics at the University of Westminster, took to X and called the video "a straightforward 1930s Germany-style cartoon; one of many violations of electoral conduct rules."
Battling the Ghosts of AI
Identifying deepfakes is imperative during the ongoing elections.
Efforts are already in place. The global social conglomerate Meta recently collaborated with the Misinformation Combat Alliance (MCA) to form a WhatsApp fact-checking helpline to aid people in recognising AI-fabricated materials and be better informed.
The director of public policy at Meta India, Shivnath Thukral, underlined the importance of the detection system, saying that it facilitates “people on our platforms with resources and tools that make it simpler for them to identify content that has been generated using AI tools and curb the spread of misinformation”.
Similarly, OpenAI, the company that gestated ChatGPT, has formulated a deepfake detector for disinformation researchers. The tool is built to detect primarily the images generated by the company's own popular image generator, DALL-E. While OpenAI claims that the detector can whistleblow 98.8 % of images created by its latest model, DALL-E 3, it acknowledges that the tool isn't designed for other prevalent generators like Midjourney and Stability.
Forensics outwitted
Despite these steadfast efforts, the sheer volume and sophistication of deepfake content continue to outpace detection capabilities. Moreover, the reliance on user reports to flag potential deepfakes means that many false or misleading videos may go unnoticed until significant damage has been done.
"The quality of deepfakes is so good that even forensic labs can take days to establish the veracity of a medium," cyber security expert Jiten Jain was quoted as saying in a media report. In addition, it is common knowledge that no AI algorithm is 100% accurate. When it comes to a subject so instrumental to electing the next leader of the country, a fallacy by even a minor estimation could be nothing short of monumental.
Public awareness and digital literacy are crucial to mitigate the impact of deepfakes but they are often lacking. Many people, especially the older and less tech-savvy generations, are not familiar with the notion of deepfakes or how to recognise them, making them more susceptible to believing and sharing false information.
As technology becomes democratised, ironically, the democratic principles of a nation could be compromised because of its misuse.
Innovation in detection technologies, enhanced public education and constant awareness, robust legal reforms, and greater international collaboration seem the only plausible guardians of the fragile truth.
Ultimately, all this casts the Indian voter into a deliberate scepticism where all information is guilty until proven authentic through rigorous scrutiny because the devil often lies in the details.
Suvrat Arora is an independent writer based in Bengaluru, India. His works have appeared in Al Jazeera, The Telegraph, Outlook India and elsewhere. He can be reached at suvratarora06@gmail.com
Continue reading with HT Premium Subscription
