Deepfakes deceive voters from India to Indonesia before elections

During his work in the film and television industry creating AI-based visual effects and voice clones, Divyendra Singh Jadoun began to receive requests from political figures wanting his technology for their campaign videos. His company, The Indian Deepfaker, experienced a surge in interest due to last year's intense local elections in his home state of Rajasthan, as well as the upcoming national elections in May. However, Jadoun, aged 30, expressed concerns, noting the alarming ease and speed with which deepfakes can be created and the difficulty for the public in discerning real from fake. He noted the absence of any guidelines surrounding deepfake technology and its potential to manipulate individuals' voting decisions.

Political deepfakes are not unique to India. Worldwide, deepfake videos are producing misinformation and confusing voters, technology experts and officials state. Notable examples have surged in India, Indonesia, Bangladesh, and Pakistan ahead of elections. India's Prime Minister Modi has expressed his concern about deepfake videos, and made it known to social media platforms that they could face the loss of their safe harbor status, a protection against liability over third-party content, if action is not taken to deal with this problem.

In Indonesia, AI-created videos featuring the presidential candidates are growing in popularity and could influence the election's outcome, shared Oklahoma State University Assistant Professor Nuurrianti Jalli. She adds that deepfakes are another tool being used to propagate misinformation and manipulate public opinion, particularly in areas where misinformation is already prevalent.

Deepfakes have also been noticeable leading up to elections in countries like New Zealand, Turkey, and Argentina. Divyendra Singh Jadoun in India has been reluctant to make deepfakes, but this reluctance is not shared by all, and misuse of this technology is growing. The U.S. non-profit Freedom House recently highlighted in their report that AI is accelerating the creation and spread of misinformation, making it cheaper and more effective.

However, in some countries such as Pakistan, the benefits and popularity of deepfakes are recognized. Recently, Imran Khan, the imprisoned former prime minister, used an AI-generated image and voice clone to address an online election rally, attracting more than 1.4 million viewers on YouTube.

Social media giants Meta, who owns Facebook, Instagram, and WhatsApp, as well as Google, owner of YouTube, vowed to remove manipulated content that could potentially mislead its users. They also require creators to openly admit any alterations or synthesis of content through labeling. Yet, despite this apparent concern, social media platforms seem to be failing in their responsibilities. As Asia policy director at advocacy group Access Now, Raman Jit Singh Chima stated, the platforms' lack of responsiveness and proactive measures is a very dangerous sign.

Back in India, Jadoun has refused deepfake video requests for local elections but is preparing them for the upcoming national elections. These videos aim to be sent to party workers, not voters, via personalized messages created by politicians. Jadoun stated that his videos have the potential to significantly affect voting behavior as they can be shared widely, but he added that he will add watermarks showing that the videos were created using AI to prevent misunderstanding.