Artificial Intelligence (AI) has entered our hands through the doors of social media. Is its impact limited to entertainment or creative work? Be it political instability or electioneering, AI is influencing us in many ways.

That is why AI has become a new topic of debate and concern for politics and elections around the world. And in recent elections around the world, the use of AI has become central to electoral strategy. From campaigning, voter engagement and image building to misinformation and voter suppression, AI has changed the very nature of the electoral process. 

Let us look at the 2024 Lok Sabha elections in India. Many have called it the ‘first AI election’. Because in Tamil Nadu, the Dravida Munnetra Kazhagam (DMK) party used an AI-created avatar of its late leader M Karunanidhi in its election campaign. Prime Minister Narendra Modi also used AI to deliver a real-time translation of a Hindi speech into Tamil. This opened up the possibility of direct dialogue with non-Hindi-speaking voters. On the other hand, the ‘deepfake videos’ circulated claiming that Bollywood actors Aamir Khan and Ranveer Singh made anti-Modi comments had garnered equal attention. 

The use of AI was also a subject of controversy in the 2024 US presidential election. Thousands of voters in New Hampshire received a robocall in the voice of an AI clone that sounded like President Joe Biden, urging them not to vote in the primary to "save their votes for November." The incident was deemed an attempt at voter suppression, and charges were filed against the consultant involved.

The direct and indirect use of AI in politics seems to be increasing. The use of AI in itself is not bad, but its misuse is being used to spread misinformation, which is a criminal act: Shailendra Raj Giri, President, AI Association Nepal Nepali politics is also close to the effects and side effects of AI. From using AI to harass leaders on social media, to spreading false information, leaders, activists, supporters, and social media influencers seem to be taking the lead. Some of them are considering this issue as light entertainment, while others are seeking political gain.

 For example, last Wednesday morning. A short video was posted from the Facebook page of Rastriya Swatantra Party (RSWP) leader and outgoing MP Asim Shah. In the eight-second video, a person who looks like RSWP President Ravi Lamichhane, Minister for Industry, Commerce and Energy Kulman Ghising and Kathmandu Metropolitan City Mayor Balen Shah are seen walking together. The ‘Falgun 21’ mentioned in the caption has also been interpreted in political circles as a sign that Ravi, Kulman and Balen may stand on the same political front in the upcoming elections. In the caption of the video posted at 8:28 am, leader Shah wrote ‘Coming Soon Together – Falgun 21’. The video has been shared more than 600 times by the time this content is prepared.

However, this viral video is not real. The video, created with the help of AI (artificial intelligence) technology, seems to be trying to convey different messages. On the one hand, the symbolic message that Ravi, Kulman and Balen are together has been conveyed, while on the other hand, some users on social media have commented that Dharan Mayor Harkaraj Rai should also be included. Some have interpreted it as a 'media trial'. 

***

This is not the only example of the use of AI-generated content in political circles. Another similar example is the video of UML leader Pradeep Gyawali crying during the UML general convention. The news that Gyawali left the UML general convention held at Bhrikutimandap on December 11 was published in some media outlets. After that, photos and videos that appeared to show him crying began to spread on social media.  In a short time, the content went viral on platforms including Facebook, X, TikTok. However, in reality, those photos and videos were not real. On 3 Mangsir, a video of him talking to media persons was modified with the help of AI technology to spread a false message. Another photo created using AI during the UML general convention also became very popular on social media. The photo, which showed the convention venue being packed with people, garnered widespread attention after it was posted by a responsible leader and worker within the UML. Later, it was confirmed that that photo was also created by AI.  Use in propaganda

Some people have also started using AI in political propaganda. For example, UML general secretary candidate Surendra Pandey has also posted AI-generated content   from his official social media account. In the 21-second video posted at 8:34 am on Tuesday, he presented the work he did as Finance Minister through AI. In the video, the photos published in the newspaper are made to move. Along with the sad background sound, the photos of Nepali Communist Party leader Pushpalal Shrestha, UML leader Madan Bhandari and Bharat Mohan Adhikari are also seen ‘moving’. Similar AI-generated video materials have been posted from an account called ‘Surendra Prasad Pandey Secretariat’.

Is it only UML that has used AI in propaganda? No. Recently, it seems that AI-generated materials have also started being used to spread propaganda for other parties and defame leaders. A video of a person who looks like Progressive Democratic Party leader Janardan Sharma speaking is found on social media, in which he seems to have made the statement, ‘I did not split the party, I broke your illusions. It is a thousand times better to become a mason who lays a new foundation than to die buried in a dilapidated house.’ Similarly, workers of the same party have also posted AI videos that look like a big propaganda rally.

AI-generated videos have also been made praising Nepali Communist Party coordinator Pushpa Kamal Dahal ‘Prachanda’. In one video, a photo of Dahal is placed on the left, while on the right, a large crowd carrying Nepali flags is seen chanting slogans like ‘Ravi, Balen, we don’t know anyone, we won’t vote for Congress, UML’. Videos praising a person who looks like Ravi Lamichhane, the president of the Nepali Congress Party, are also going viral. In one video, a person who looks like Ravi is seen walking with a white tiger. In the background, an audio that sounds like Ravi’s voice is playing – ‘I am a man of the ground, I can stoop to the ground… People who commit crimes must be scared by media headlines. The Nepali Congress Party is not a party of cowards, Ravi Lamichhane is not a leader of cowards.’ Excuses too  AI seems to be starting to be used not only for propaganda, but also as an excuse to hide mistakes. During the protest on Bhadra 24, a video of money burning in a fire at the residence of Nepali Congress President Sher Bahadur Deuba was released. Deuba's secretariat denied the video, claiming that it was created by AI. After the

, Deuba also tried to distance himself by saying that all those videos were created by AI. However, it has been confirmed that those videos were not created by AI. Kantipur had also done a fact check on this. Before that, when the audio of the then Communications Minister Gokul Prasad Baskota regarding a Rs 700 million transaction was made public in 2076, he had denied it, saying that it was created by AI. However, after widespread criticism, he resigned on Falgun 8, 2076 on ethical grounds.

Defaming the old, praising the new

Most of the AI-generated content that goes viral on social media is of a nature to defame the leaders of old parties. Deepfake videos showing UML Chairman KP Sharma Oli and NCP Coordinator Dahal fighting each other, showing Congress President Deuba, Oli and Dahal being arrested by the police while sitting together, and showing Congress General Secretary Gagan Thapa grazing goats are all over social media.

On the contrary, there is more content praising the so-called new or alternative leaders. Videos showing Balen Shah and Oli fighting, scenes giving the message that Balen won and Oli was defeated, and videos showing Ravi Lamichhane ringing a bell at Oli have also been widely shared.

Signs of increasing AI in elections

AI Association Nepal President Shailendra Raj Giri says that the direct and indirect use of AI in politics is increasing. ‘The use of AI in itself is not bad, but its misuse is being done to spread misinformation, which is a criminal act,’ he says. As the election approaches, such use will increase, says Santosh Sigdel, an advocate and executive director of Digital Rights Nepal. “Political parties, groups supporting them, or activists can use AI content to create an atmosphere in their favor or to harm the opposition,” he says. “Groups that do not want the election or want to disrupt the election can also use such content.”

According to journalist and fact-checker Umesh Shrestha ‘Salokya’, it has now become very easy to create propaganda content using AI. “A deepfake video can be created from a single photo. This is a good aspect of technology, but its misuse is a serious problem,” he says.

Similarly, journalist and fact-checker Deepak Adhikari also says that identifying AI-generated content is becoming increasingly complex. “Earlier, AI content could be easily distinguished. Now, as technology improves, it is becoming increasingly difficult for users to distinguish between real and artificial,” he says.

Lack of regulation

AI Association President Giri also says that using AI with wrong intentions is illegal. ‘Using someone’s photo or voice for misleading purposes is a crime under cyber law,’ he says. But the draft code of conduct made public by the Election Commission does not have a clear provision on the use of AI. Although propaganda, misleading information and hate speech are prohibited, there is no direct mention of AI-generated content.

Considering this as a serious weakness, advocate Sigdel says, ‘The Election Commission must formulate a clear policy on the use of AI. A system is needed to distinguish which use is legal and which is illegal.’ According to him, the commission should coordinate with ‘Big Tech’ companies and develop a mechanism to identify and control misleading AI content during elections. Journalist and fact-checker Shrestha also emphasized that the Election Commission should clarify the use of AI in the code of conduct. 

The increasing use of AI in politics seems to be adding challenges not only to the form of information flow, but also to the democratic process. How deep its impact will be in the upcoming elections, its regulation and accountability are still unclear. The use of AI is sometimes a matter of controversy, sometimes an election strategy. Sometimes it is being used to incite voters, sometimes to make voters more informed. 

The campaign of former Prime Minister Imran Khan in Pakistan’s 2024 election became the most famous example of the use of AI. Although Khan, who is in prison and banned from public appearances, was unable to make a live speech, his party Pakistan Tehreek-e-Insaf (PTI) ran a virtual campaign with the help of AI. The ‘victory speech’ released after the election was created by AI. The video, which presented a text written in prison in Khan’s own artificial voice, used his old scenes. 

A similar exercise took place in South Korea in 2022. There, a digital avatar called ‘AI Yun’ was released. This avatar of candidate Yun Suk-yeol was constantly interacting with voters online, using a youthful language style and humor. Meanwhile, in Slovakia, AI-generated audio leaked at the last minute before the 2023 election affected the election. The rival candidate was defeated because he did not have time to refute the audio leaked during the election silence period.

Along with these developments, various studies have expressed concerns about the use of AI in future elections. A study conducted by Elon University Poll and the Imagining the Digital Future Center in May 2024 concluded that the use of AI in elections will affect the election.

According to the study, 78 percent of Americans believe that AI will be used to influence the election results, while 73 percent are concerned that AI will be misused to spread fake information on social media. It has also been seen that most citizens themselves are unable to distinguish fake content created by AI. 52 percent said they cannot recognize fake audio, 47 percent said they cannot recognize fake videos, and 45 percent said they cannot recognize fake photos.