Deepfakes and Democracy

As Nepal heads toward the March 5 House of Representatives (HoR) election, the country’s democratic process is facing a challenge it has never encountered at this scale: the rapid spread of deepfakes, fake news and AI-generated misinformation. What was once dismissed as harmless online mischief has now become a serious threat to election integrity, public trust and social stability. A recent viral video involving Balendra Shah (Balen), a HoR candidate from Jhapa constituency no. 5, captures the gravity of the problem. The clip, showing Shah aggressively pushing his own party cadres in Damak, first drew praise from supporters who saw it as a display of assertive leadership. Soon after, criticism followed, with many questioning his temperament. When Shah’s team claimed the video was a deepfake, the controversy only deepened. Even days later, AI experts have been unable to conclusively determine whether the video is manipulated or authentic. The damage, however, was already done. This episode is not an isolated one. Another widely shared image showed the coordinator of the Nepali Communist Party Pushpa Kamal Dahal appearing to cry while holding a photograph of a “martyr” from the decade-long civil war. For digitally savvy users, the image was an obvious fabrication. But for many others—particularly older citizens or those with limited digital literacy—it appeared genuine, evoking sympathy or outrage. From fabricated images of leaders crying, dancing or begging for votes to AI-generated videos, audio clips and even songs filled with inflammatory political messages, Nepal’s social media space has been flooded with misleading and often toxic content. This is likely to grow further as the election date nears further.