Can Deepfake Jeopardize Elections 2024? Startups Come Up With Coping Mechanism The USA, Nigeria, and Bangladesh are a few that have been impacted by deepfakes during elections. The 2023 State of Deepfakes Report ranks India as the sixth most vulnerable country to deepfake

Ten months ago, Senthil Nayagam took to LinkedIn to share about his artificial intelligence startup, "Muonium Ai is building capabilities for small creators like YouTubers and Instagrammers to create movie quality content at fraction of the cost and time of traditional Movie Making, CGI, VFX, Post Production process." In 2024, he was the brain behind the resurrection of M Karunanidhi before a live audience on a large projected screen. It was done to congratulate his 82-year-old friend and fellow politician TR Baalu on his autobiographical book launch and praise the leadership of MK Stalin, his son and the current leader of Tamil Nadu.

"I didn't anticipate this thing to absolutely blow up," shared Nayagam in the media. As Elections 2024 are underway, political parties and members are gearing up to leverage AI to boost the voter bank. This election will go down in history for being heavily influenced by artificial intelligence, particularly deepfakes.

According to the '2023 State of Deepfakes Report' by 'Home Security Heroes,' a US-based web security services company, there were over 95,000 deepfakes online in 2023, up 550 per cent since 2019. The USA, Nigeria, and Bangladesh are a few who have been impacted by deepfakes during elections. The report further ranks India as the sixth most vulnerable country to deepfake.

Backstory of Tech-powered propaganda

A first-of-its-kind usage in the Indian landscape was done by BJP and its MP, Manoj Tiwari, four years ago. The original clip saw Tiwari criticizing Arvind Kejriwal and encouraging voters to vote for the BJP in Hindi. The altered video saw speaking in Haryanvi, the dialect spoken by the target voters of BJP. The party told Vice, a Canadian-American magazine, that the video reached approximately 15 million people in 5,800 WhatsApp groups.

In April, just a few days before the elections, Bollywood actors Aamir Khan and Ranveer Singh could be seen criticizing the BJP and PM Modi and asking the public to vote for Congress. Singh was seen criticizing the BJP and the PM, "The purpose of Modi ji was to celebrate our miserable life and fear, our unemployment and unemployment and we are moving towards the era of injustice."

These videos were fake, yet Singh's video alone garnered over 438,000 views, 2,900 reshares, and 8,700 likes when shared by Congress spokesperson Sujata Paul on X. While Paul knew the video was not real, she did not delete "It has creativity for sure".

Other cases include KT Rama Rao, of Bharat Rashtra Samiti, calling on people to vote in favour of the Congress; and Congress leader Vijay Vasanth's team creating the late father's two-minute AI video.

PM Modi's speech at Kashi Tamil Sangamam last year at Varanasi was translated in real-time into Tamil for the attendees, "Those from Tamil Nadu, I request them to use your earphones (to listen to the speech) using AI technology for the first time."

Identifying and working towards deepfake

The deepfake videos are emerging as a matter of concern for the affected parties and general citizens. "The election season is on and social media buzz every day with new news, statements, images, and videos circulating about political parties or individuals. The public should be aware and not fall for AI-generated images, deep fakes, or fraud accounts peddling misinformation," shares Amit Relan, Co-Founder and CEO, mFilterIt.

To combat mass exploitation, tech giants such as Adobe, Amazon, Google, IBM, Meta, Microsoft, OpenAI, TikTok, and Elon Musk's X have joined forces to combat AI-generated election interference. As many as 60 countries will be entering the election mode in 2024. Meta, weeks leading up to the Indian elections, announced it would be empowering its fact-checkers to label content generated with the help of artificial intelligence (AI) or deepfake technology as "altered" on its platforms Facebook and Instagram. Furthermore, such content would appear lower on the user feeds.

"To ensure public safety, companies must take responsibility and implement measures to combat deepfakes and disinformation. This includes investing in advanced detection technologies to identify and flag deepfake content, as well as collaborating with experts to develop effective debunking methods," shares Ivana Bartoletti, Global Chief Privacy & AI Governance Officer, Wipro.

AI is being used to spread disinformation, create deep fake elections; flood voters with highly personalised propaganda, and translate speech into regional languages for microtargeting in campaigns.

Ranjan R Reddy, Founder & CEO, Bureau feels that AI is a double-edged sword, "While AI fuels sophisticated deepfakes that exploit trust and security gaps, it also offers powerful tools to combat them. Generative AI can analyze vast datasets to identify patterns of fraudulent behaviour."

Startups combating malefic content

"The election season is on and social media buzz every day with new news, statements, images, and videos circulating about political parties or individuals. The public should be aware and not fall for AI-generated images, deep fakes, or fraud accounts peddling misinformation," shares Relan. The startup scans across social media, the web, and the dark web to identify any brand infringement and misuse of logo, theme, or names using 'keyword' based search, and open-source intelligence, AI/ML driven tech.

Kroop.ai, an ethical synthetic data solution platform, enables users with its Deepfake Detector to detect whether an image, video, or voice is AI-fabricated, AI-generated or real. Group Cyber ID is an Indian startup that offers digital forensic solutions for images, videos, and audio, among other digital components. Its services include advanced cyber security, IT audit, digital forensics, and threat intelligence services.

Staqu Technologies, an AI-based security and big data analytics service provider, is also working on a deepfake model. "We generate fake photos and then we train the Al GAN model, where we try to make it so strong that it does not make any error in identifying real versus fake," said Atul Rai, co-founder and chief executive officer, Staqu Technologies.

Tools and software such as Google Fact Check Explorer, VerifyThis, or Snopes can be used for deepfake detection.

What can be done? Public Role

According to a study conducted by iProov, 71 per cent of respondents globally do not know what a deepfake is, while 43 per cent admit they would not be able to detect one.

MeitY, in collaboration with several stakeholders, in November commented on the pressing need to regulate deepfakes and introduce new rules, enact a new law, or amend existing regulations. "We agreed that we will start drafting the regulation today. And within a very short timeframe, we will have a new set of regulations for deepfakes," said Minister Ashwini Vaishnaw. The advisory stated that social media platforms should effectively communicate with users about the need to avoid posting or spreading deepfakes, as they can face punishment under the Indian Penal Code.

Bartoletti feels that promoting media literacy and critical thinking among the public is crucial, "By providing educational resources and raising awareness about deepfakes and their dangers, individuals can become more discerning consumers of information, less susceptible to manipulation."

The general public can use Intel's Real-Time Deepfake Detector; Sentinel; DeepWare AI; Sensity AI; and Microsoft's Video Authenticator Tool to do their due diligence before solidifying opinions and shifting stances.