Recently, an edited video of Home Minister Amit Shah went viral, falsely showing him promising to abolish reservations for Scheduled Tribes, Scheduled Castes and Other Backward Classes. A BOOM fact check found that the video was doctored using video editing tools, but was incorrectly labeled as a “deepfake” by other Bharatiya Janata Party leaders and various mainstream media outlets. .
Indian political parties are experimenting with AI in various ways this election season. For example, her AI voice clones of politicians are used to craft messages to executives and voters. Political parties are also circumventing social media rules by using satirical videos that use voice cloning, face swapping, and other AI editing techniques to target political opponents. These videos will be posted on the official Instagram and YouTube handles.
However, blatantly deceptive and harmful deepfakes are being shared by IT cell workers or through agents and spreaders supporting the party and its ideology.
Misinformation peaks in India during the election season, and so far shallow fakes and cheap fakes continue to make up the bulk of the misinformation seen online.
Not necessarily a deepfake
Several mainstream media outlets, including The Indian Express, Times Now, Republic, and DNA, called the doctored Amit Shah video a deepfake. Prime Minister Narendra Modi, speaking at a rally in Satara, Maharashtra, also mentioned that the video was altered using artificial intelligence.
However, BOOM's analysis of the video shows that various parts of the speech were spliced together to take Shah's statement on abolition of Muslim reservations in Telangana out of context ahead of the 2023 Telangana Assembly elections. It was also discovered that the documents had been falsified. Our analysis confirms that this video is not a “deepfake.” This means it has not been modified or created using artificial intelligence or deep learning algorithms.
Last November, while addressing the media at Diwali Milan hosted by the Bharatiya Janata Party at its national headquarters in New Delhi, Prime Minister Modi highlighted the threat of “deepfakes.” As an example, he went on to point out a video of himself allegedly dancing garba and called it a deepfake.
BOOM fact-checked the video and found that it is not a deepfake or an edit, but instead shows a real video of a Narendra Modi lookalike named Vikas Mahante dancing garba during Diwali festival in the UK.
Days before the first phase of elections, a video of Dinesh Lal Yadav, the BJP candidate from Azamgarh, went viral. There, he was seen talking about how PM Modi and UP CM Yogi Adityanath want to avoid having children to eliminate unemployment. After the video was widely shared by Congress supporters, BJP IT Cell chief Amit Malviya tweeted that the video was a deepfake.
BOOM analyzed the video using deepfake detection tools and also obtained the original video file from the reporter who shot the video, confirming that the video is real and not a deepfake. Although his comments were rearranged in the viral video, it was not a deepfake and did not change the meaning of what he said.
“We should absolutely expect the term 'deepfake' to be misused, just as terms like 'misinformation' and 'fake news' are used by politicians to ignore evidence they don't like. “It's the same as before,” said Prateek Wagre, executive director of Internet. Freedom Foundation.
In an interview with BOOM, Wagle highlighted two different types of cases where the term “deepfake” has been misused. “For one, there may be a lack of awareness of the nuances of the difference between an edited version, a 'cheapfake' and a deepfake. The other is to call it a deepfake and completely ignore the evidence. That’s what I’m doing.”
real political deepfakes
The use of deepfakes in politics is not new. The first use of such technology in an election context was seen on February 7, 2020, a day before Delhi headed to the assembly polling booth.
Several videos have been released showing Bharatiya Janata Party leader Manoj Tiwari criticizing the Aam Aadmi Party government and its leader Arvind Kejriwal and urging people to vote for the Bharatiya Janata Party. These videos show Tiwari speaking in English, Hindi and Haryana's Hindi dialect.
However, Vice discovered that only the Hindi version was originally shot by Tiwari. The English and Haryanvi versions were actually fabricated using a 'lip-sync' deepfake algorithm trained on videos of Tiwari's speeches.
Neither the Election Commission of India nor the BJP has publicly commented on Tiwari's deepfake video.
Most recently, BOOM used artificial intelligence to fact-check multiple videos that were created or modified and shared in the context of an ongoing poll.
Less than a week before voting begins, a video of Congress leader Rahul Gandhi has surfaced on social media in which he can be heard announcing his resignation from the party. BOOM found that the video actually showed Gandhi filing his election candidacy from Wayanad, Kerala, and was overlaid with an AI voice clone.
A few days before the elections, two videos of Bollywood actors Aamir Khan and Ranveer Singh criticizing the Bharatiya Janata Party-led government went viral online. BOOM analyzed both of these videos and found evidence that they had been altered using AI voice cloning. Click here and here to read the fact check.
Another video was posted on social media between the first and second phases of the election. There, Congress leader Kamal Nath could be heard promising Muslims land to build a mosque and Article 370 being reinstated. BOOM discovered that this video had also been altered. Eggplant's voice is replaced by a replica of the AI's voice clone.
The Election Commission of India is yet to issue a statement acknowledging the emergence of such AI-driven disinformation related to polls. Even before India's elections, deepfakes were circulated during polls in neighboring Pakistan and Bangladesh earlier this year.
How should deepfakes be regulated?
Regulation of AI has become a hot topic in India after an attempt at a recommendation by Electronics and Information Technology Minister Rajeev Chandrashekhar failed, leading to massive confusion and multiple clarifications from the minister. .
“Whether we use the term ‘deepfake’ or not, whether we look at the broader issue of ‘synthetic media’, we need to be clear about the concepts we are trying to regulate. “It's becoming increasingly obscured by debates about it,” Wagle points out.
The novelty of AI tools has left many confused as to what exactly is being discussed, and the deliberate misuse of the term “deepfake” is expected to make the situation even worse. So how can we curb the misuse of such technology?
Radhika Jhalani, volunteer legal advisor at SFLC.in, referred to the now-withdrawn recommendation by MeitY, highlighting the evolving nature of technology that “can change on a daily basis” and providing a balanced approach to new laws. approach is recommended. 2024 is a major election year globally, and deepfakes are a cause for concern. The laws that are implemented need to balance the misuse of technology and freedom of speech,” Radhika told BOOM.
Wagre believes that companies should avoid rushing to new regulations and instead look at existing laws to find regulatory gaps.
“What are the different ways in which it (deepfakes) is being used in harmful situations? And what does existing law say about counterfeiting, identity theft, etc.? And how are laws currently inadequate? “We need new laws to cover them, but do we need to amend the laws or do we need to improve the redress mechanisms?” added.
