Artificial intelligence in the elections: how the technologies influenced democracy in 2024

Photo: Shutterstock

According to Global Risks Report 2024, disinformation and manipulative campaigns have become the biggest global threat in the short term perspective. At the same time, in the long-term forecast for the next decade, disinformation remains among the key challenges, ranking fifth in the global risk ranking.

Photo: World Economic Forum

At the beginning of 2024 World Economic Forum published a report, that displayed the forecasted level of disinformation risk in different countries — from India, where disinformation was recognised as the No. 1 threat, to the United States and Europe, where it was among the top ten most critical risks.

Photo: World Economic Forum

Now, reflecting back on the events of 2024, we see how accurate these predictions were. Elections were held in more than 60 countries around the world - it was a record year for democracy. But it was also a record year for disinformation.

From the United States to Indonesia, candidates, their campaigns, and even ordinary supporters used AI to influence voters, manipulate emotions, spread fake news, or... deny the truth.

In this article, we have collected the most striking examples of how artificial intelligence made or broke the further faiths of candidates in their quest — to gain the power around the world.

We will talk about the US, India, Japan, Indonesia, and Pakistan.

Illusion of endorsement: how AI frames the image of a candidate in the minds of voters

In August 2024, a fierce battle was being fought not only for votes, but also for... reality. In Detroit, more than 15 000 people came out to support Vice President Kamala Harris and her running mate, Tim Walz. Photos and videos from the demonstration flooded the media. But Donald Trump said the crowd was a fiction generated by artificial intelligence. Without any evidence, without refutation after debunking.

Illustration of a scene from the demonstration of K. Harris and T. Walz in Detroit on 7 August. Photo from open sources

Illustration of D. Trump's post. Photo from open sources

This is a classic example of the so-called ‘liar's dividend’ — when the very existence of technologies such as deepfakes or generative AI becomes a tool for denying the truth. The term ‘liar's dividend’ was first described in detail in a study in 2018. It is a situation where politicians and public figures use the growing distrust of digital content to deny truthful information. If a video or audio exposes them to corruption, lies or crimes, it is enough to say that it was artificial intelligence imitating a speaker's voice or face. A society overwhelmed by fakes, doubts and information noise is tempted to believe this.

A study was conducted in the United States with the participation of 15,000 people. The results showed that politicians who declare the truth to be ‘fake’ receive more support than those who simply remain silent or apologise.

During the 2024 elections, disinformation tactics in the United States underwent a transformation: while in 2016 the main threat was external actors, this time fakes were created and spread by American voters themselves. Conservative bloggers actively used AI to post images of Donald Trump surrounded by black voters, creating the illusion of his massive support among this electorate, a key element that ensured Joe Biden's victory in 2020.

The image, which was created and spread by radio host Mark Kaye and his team using AI, is one of dozens of fakes depicting black Trump supporters. AI-Generated image

One of the most obvious examples is an image of Trump with a group of black voters on a porch. Initially, it was a satirical post, but the situation changed when user X, who goes by the nickname Shaggy, added a caption claiming that Donald Trump had stopped his motorcade to meet these people in person. The result — more than 1.3 million views.

The image was widely shared on social media with the caption that Trump had stopped his motorcade to take a picture with these men. AI-Generated image

This is just one case that demonstrates how Trump's extensive network of influencers works. They have no official status, but their messages fit perfectly into the candidate's favourable information background. Thanks to this decentralised model of disinformation, each fake is perceived as ‘people's opinion’ rather than a planned campaign, making it even more difficult to deal with manipulation.

Despite the efforts of social media to label artificially created images and the work of fact-checkers, the wave of fake content continues to frame public opinion.

«Even if you know it's not true, the constant repetition of fake news that exaggerates support for a particular candidate can leave a mark inside your mind», — says Peter Adams, senior vice president of research at the News Literacy Project.

Trump and his network of loyal influencers actively used fake content during the election campaign. Among the examples— the publlication of an image of Taylor Swift allegedly supporting Trump, when in fact the singer openly favoured Kamala Harris.
Illustration of fake news about Taylor Swift's alleged support for Trump

Another AI-generated image showed Trump at the epicentre of a flood after a hurricane on the eve of the election, walking alongside rescuers:

AI-Generated image

And another photo showed him handing out cheques for $1,300 to victims.

AI-Generated image

All these stories had nothing to do with reality.

These are just a few examples from a large campaign where a network of influencers close to Trump massively distributed AI-generated fakes in support of him. People scrolling through their news feeds rarely think about the authenticity of what they see - most don't even notice the fake news. However, it is the frequency and the volume of such content that creates a certain image of a candidate in the minds of the public, regardless of whether it is true or not.

When politicians speak from the dead: how AI has changed elections in India

The 2024 general election in India was not only the largest in the world in terms of the number of voters - almost 1 billion citizens - but also became a telling example of how artificial intelligence can change the very essence of political communication.

In a country where the literacy rate is only 77,7%, and more than 22 per cent of citizens have difficulty reading and writing, traditional methods of campaigning have proven insufficient. Political parties, including the ruling BJP, relied on innovations: voice calls, audio messages, personalised video messages, massive dipfakes and even the ‘resurrection’ of past leaders.

Robotic personalised calls became the key tool. The cost of such calls was 8 times cheaper than the regular cost of operating call centres, but they created the effect of personal contact. Millions of Indians received a call with Narendra Modi's voice calling for people to come to the polls. For many people, this was an event - after voting, people shared that ‘the prime minister himself invited them’. Some voters didn't even realise that they were talking to an algorithm, not a person.

AI helped to overcome the language barrier — in a country with 22 official languages and hundreds of dialects, Modi ‘spoke’ in dialects he had never known.

However, personalisation was not the only technological innovation. The election campaign turned into a war of deepfakes. Leaders ‘danced’ in fake videos, sang Bollywood songs, and made fictional statements. But the real know-how came from the ‘resurrected’ politicians.

There are at least three known cases of parties using AI to create messages from long-dead leaders. Virtual А. I. Karunanidi, an iconic figure in Tamil Nadu, encouraged people to support youth and democracy. The voice of Jayalalithaa, who died in 2016, addressed citizens with a request to support a particular candidate. In another case, a politician’s son “resurected” his father to campaign in his own favour.AI-Generated image

Japan chooses the future: how a programmer with an AI avatar changed Tokyo's political communication

In a country where tradition is intertwined with high technology, a political campaign of 33-year-old programmer Takahiro is remembered as a breakthrough in the interaction between the government and citizens. In 2024, this entrepreneur, inspired by his own sci-fi thoughts on the role of artificial intelligence in politics, decided to challenge the established rules. His goal was simple and ambitious at the same time - to break the outdated model of one-way communication between candidates and voters.

With no political experience, but with a strong knowledge of technology, Takahiro launched his campaign for the post of Tokyo governor. Within a few months, he was ranked fifth among 56 candidates, attracting the attention of not only the Japanese but also experts from around the world.

Takahiro's campaign has become a model of how artificial intelligence can be integrated into democratic processes in three steps:

From listening to comprehension.
The first step was a large-scale public opinion poll. Through comments on social media, AI analysed thousands of responses, grouped them into thematic clusters, and formed a real picture of the requests and problems of Tokyo residents.
From manifesto to dialogue.
Instead of dictating a ready-made programme to voters, Takahiro opened a platform for suggestions. His team collected and systematised the ideas of citizens, turning their opinions into the basis of a political course.
AI in touch.

The culmination was the launch of an avatar - AI Takahiro. The virtual candidate answered voters' questions in real time, explained his positions, and demonstrated a new level of accessibility of the politician. «For the first time, people felt that they were not just heard, but listened to and taken into account», — Takahiro stated in an interview.

Although he failed to win, in February 2025, it became known: the current governor of Tokyo, his former opponent, invited Takahiro to cooperate. Together, they launched a project in which Takahiro's public idea collection system became an official tool of city governance. Now, artificial intelligence helps the authorities identify the voices of citizens that deserve attention - not through bureaucratic procedures, but through a live digital dialogue.

Artificial intelligence in the service of politics: how Pakistan became an arena for diplomatic facilitation and virtual speeches

In 2024, when Pakistan plunged once again into a whirlpool of political instability, it was not so much the election result that attracted the world's attention as the way the opposition fought for its voice. In a country where democracy traditionally borders with repression and military leverage is felt at every stage of the political process, artificial intelligence has suddenly become a tool of resistance

Former prime minister Imran Khan, jailed for 10 years for corruption and banned from participating in elections, found a way to remain present in the campaign — even from behind the bars. His team launched a series of messages in which Khan's voice and image were recreated using AI. Algorithms turned notes transmitted from prison into speeches that instantly spread across digital channels. AI-Generated image

When the vote count ended on 8 February and it became known that independent candidates affiliated with Khan's Pakistan Tehreek-e-Insaf (PTI) had won the largest number of seats in parliament, it was Imran Khan's virtual voice that announced the victory. It was a symbolic culmination of a campaign where a politician physically isolated from society continued to inspire his supporters through technology.

However, the political reality of Pakistan proved to be more complex than the arithmetic of election protocols. Due to the PTI's ban, candidates were forced to run as independents, and after the election, they joined the Sunni Ittehad Council (SIC) party, which allowed them to form the largest faction in the National Assembly.

However, no single force gained an absolute majority. The power returned to the hands of familiar players: The Pakistan Muslim League (Nawaz) (PML-N), led by Shahbaz Sharif, entered into a coalition with the Pakistan People's Party (PPP) and other allies. Sharif, who had already served as prime minister, returned to the leadership of the country.

However, the Pakistani reality also showed the dark side of new technologies. In a country where patriarchal norms still define the social framework, sexualised deepfakes have become a powerful weapon against women in politics.

One of the opposition leaders was the victim of discrediting campaign — with fabricated videos of an intimate nature appearing online to destroy her reputation. In an environment where the concept of ‘honour’ is of utmost importance, and where hundreds of honour killings are reported every year, such fakes become not only a tool of political pressure, but also a real threat to life. Pakistan is experiencing rapid growth in internet access. However, this digital breakthrough is accompanied by a low level of media literacy. For millions of new users, everything that appears on a smartphone screen is perceived as true. In such an environment, dipfakes have a special power of influence - society has not yet learnt to doubt what it sees.

Indonesia: From ‘bloody general’ to cute picture for TikTok generation

In February 2025, Indonesia became another victim of a dangerous game with reality, where the line between real and fictional is blurred by technology. This time, Deepfake was used not by politicians but by fraudsters. They created a video in which the president of the country personally ‘addressed’ citizens with a request to... transfer money.

In the fabricated video, the president asked people to contact him via WhatsApp and pay from 250,000 to 1 million rupees (about $15-$60), allegedly as an administrative fee to receive financial assistance. Of course, the victims did not receive any assistance. This case has become a symbol of a new era of fraud, where phone calls from ‘pseudo-bankers’ are no longer necessary - a convincing video with the face of the leader of the state is enough.

It is interesting to note that the Indonesian president himself had actively used the power of visual manipulation a year earlier, but in his own political campaign. Known among the opposition as the ‘bloody general’ due to accusations of killing activists, he had a negative image among the older generation and human rights activists.

However, his team found a solution: they created a new visual representation — a cute, smiling ‘’baby‘’ specially adapted for the younger audience on social media. Thanks to the right targeting and the use of simple, friendly visuals, the campaign turned a tough military man into a ‘friend for young people’. The result was not long in coming - the president won, gaining the support of a generation that was unaware of his controversial past.

Indonesia has shown two sides of the coin in the use of modern technology:

On the one hand, political rebranding through an AI-generated image helped win the election by ‘’cleansing ‘’ the candidate's reputation.
On the other hand, the same trust in visual content turned into massive fraud when citizens voluntarily sent money to the ‘president’ from a deepfake video.

In a country that is rapidly digitalising, but has not yet developed critical thinking skills in relation to media, such cases are becoming a dangerous norm.

Artificial intelligence and elections: the future that has already arrived

Elections around the world in 2024 have become an unprecedented testing ground for artificial intelligence. From India to Japan, Pakistan to Indonesia, AI has been actively transforming political campaigns, ways of communicating with voters, and tools for manipulating public opinion.

This review covers only a small part of global cases, but even they allow us to identify key areas where AI has already dramatically changed the approach to electoral processes.

Main trends in the use of AI in elections:

Personalised calls and messages.
AI has allowed politicians to address millions of voters in a ‘personal dialogue’ format. Automated calls, voice messages, and candidate avatars have become the norm. The example of India has shown how effective such tools are in mobilising the electorate in multilingual societies with low levels of traditional political participation.
Deepfakes: a political weapon and a discrediting tool.
From virtual speeches by imprisoned leaders in Pakistan to sexualised attacks on female politicians, AI-generated videos and images have become a universal tool for influencing people. The trend of using deepfakes to undermine reputations, particularly in patriarchal societies with low media literacy, is especially worrying.
Spread through networks of influencers.

In the United States and other countries, fake news and manipulative content are increasingly being spread through bloggers and loyal influencers. This creates a situation where candidates are formally ‘out of the equation’ and disinformation spreads without any control, taking advantage of the audience's trust in familiar faces on social media.

Artificial intelligence is no longer just a technological novelty in election campaigns — it has become a full-fledged driver. From the creation of candidate avatars to the mass distribution of deepfakes, new tools are already changing the way we vote, who we elect, and how we understand politics.

The future of democracy depends on societies being able to recognise the scale of the challenges in time and find a balance between technological progress and the protection of the truth. Because when AI gets out of control, elections are no longer about choices at all.