Artificial Intelligence to Change Elections in India & Worldwide

Indian Prime Minister Narendra Modi recently warned against the misuse of artificial intelligence (AI) in elections in India.

His statement came shortly after a video of him dancing surfaced. Modi suspected it was a deepfake, meaning a video generated using AI that appeared real. Some fact-checkers revealed that the garba, a traditional Gujarati dance performance, was done by a Modi lookalike and was not an AI-generated clip.

Yet the video revealed the threats posed by AI in elections primarily because of the user-friendliness of myriad new apps from the generative AI stable as well as the difficulty of distinguishing the fake from real. But until Modi brought it up last week, neither the media nor the government thought the issue important enough to flag until now.

True, generative AI, the advanced branch of AI that can easily generate text and image based on prompts, is a boon in multiple sectors such as customer care, healthcare, food and beverages, and even in education. But chances of its abuse and consequences raise concerns all stakeholders, including the people, government and technologists working in new frontiers of AI.

Let’s not forget that Geoffrey Hinton, often called the godfather of AI, quit Google warning about the growing dangers. Several other top-notch figures in AI, including Elon Musk, signed a letter early this year warning of potential risks and appealed for pressing the pause button.

More worrying is the realisation that apps to verify fake videos aren’t effective enough yet and that the world is heading for elections in countries such as the US, India, Indonesia, Mexico and the UK over the next one and a half years.

If it was a false alarm for Modi in India, in the US, a deepfake of former American president Donald Trump was used in the campaign by his Republican rival Ron DeSantis.

Fauci is unpopular among Republican and right-wing voters, who account for Trump's core supporters.

Since last November, after US-based AI research and deployment company OpenAI launched its wonder app named ChatGPT, which helps create texts, audio, and video for online users with tremendous ease with prompts, the world of technology has changed.

The same company had launched DALL-E 2, a popular AI art generator, and Whisper, an automatic speech recognition system. AI technology has advanced rapidly since then with more players, including leviathans in IT, scrambling to keep pace with the changes.

A straightforward Google search for cloning voices and creating deepfakes, including nude videos and photos, generates a plethora of choices. One no longer needs the dark net for such morally questionable work.

In most parts of the world, various versions of "nudify" apps (to make nudes of people from their day-to-day, well-dressed videos) are available even on app stores of mobile and PC operating systems. Midjourney, Stable Diffusion and ElevenLabs are a few such names. There are numerous others to manipulate photos and videos.

In India, as is the case elsewhere, there have been cases of blackmail using deepfakes. Sextortion is a grave challenge that cybersecurity agencies and experts have to battle on a daily basis across the world, says Ritesh Bhatia, Mumbai-based founder of a cybersecurity firm. Tellingly, a 2019 study found that 96% of deepfakes are pornography, and 99% target women.

Four years on, it appears that the use of deepfakes has the potential to not only threaten individual privacy but even influence elections around the world.

In the recently concluded elections in Argentina, AI was widely used, setting in motion a trend that is expected to be zealously pursued in other countries too. It was equally used and misused as candidates vied with each other to reach out to their voters using generative AI.

A famous campaign by right-wing President-elect Javier Gerardo Milei used AI-generated images portraying his opponent Sergio Tomás Massa as a lion carrying a duck wearing an Argentine flag (presumably suggesting that the country is a sitting duck in his rival's hands).

Clearly, politicians and their parties are using new technologies to stay ahead of the race. In India, in the elections of 2014, Facebook and Twitter dominated the poll season. Modi, as a contender for the top post, was at the forefront of the use of social media to bypass the media to make public outreach. His grouse then was that the media was unfair to him.

By 2019, the whole world changed and along with Facebook and Twitter, WhatsApp, essentially a messaging platform, became central to the election campaign because it offered relative anonymity and larger reach as political parties competed with one another to tap the rising use of smartphones in the country.

India surpassed the United States in the number of smartphone users as early as 2017. According to some reports, around 223 million people in the US used a smartphone by 2017, compared to 340 million in India.

However, in terms of smartphone penetration rate, India reached close to 71% by 2023, which the US already reached in 2019. For a developing country, Internet and smartphone adoption in India continues to be phenomenally high.

India has many firsts to its credit in terms of our politicians making use of the latest technologies of the time. Prime Minister Modi was perhaps the first politician to use holograms for an election campaign – which he did when he was chief minister of Gujarat and then for the 2014 general election.

In my book, War Room: The People, Tactics and Technology Behind Narendra Modi's 2014 Win, I wrote: "The 3-D hologram campaign was a major coup. It had been tried out earlier in the 2012 Gujarat assembly elections. The idea came from cinematographer U K Senthil Kumar after his chance meeting with the Gujarat chief minister before the 2012 polls.

The use and misuse of technologies went hand-in-hand in India. For instance, when lynchings of a few members of the minority community for allegedly smuggling cow meat took place in parts of northern India, the government attacked WhatsApp for enabling such crimes. I reported on it at the time.

I noted that the American company had put in place controls to make mass sharing of dangerous content through its communication app cumbersome. It has also put out appeals across TV, radio, and print, asking people to help combat hoaxes.

Besides setting up an Indian arm of the company and appointing a grievance officer, WhatsApp also set a limit on forwarding messages. This measure, originally meant for India, was later adopted at a global scale by WhatsApp. Those guidelines have changed since then.

Still, it is significant that it was political trends in India that set corporations on an overdrive to combat the menace of abuse of their technologies. Alongside, social-media giants and media groups collaborated to check for fake news and content damaging to groups and individuals.

It quoted Vice to state that "the BJP partnered with political communications firm The Ideaz Factory to create deepfakes that let it target voters across the over 20 different languages used in India. The party told Vice that the Tiwari deepfake reached approximately 15 million people in 5,800 WhatsApp groups."

It is possible that there are other occasions when politicians may have used AI for running ads, but Tiwari's campaign was widely talked about although there was nothing awkward or damaging to others in those videos.

We are now spoilt for choices, and unlike earlier when experts were required to make deep fakes or similar AI-generated content, various new players have entered the scene, a few being Deepswap, Lensa AI, Deepfakes Web, and so on. Premium versions of these apps offer quick results. There are also companies in India that make fake videos and deepfakes on demand.

Within India, there are those who use AI to cut through the maze of political campaigns by established players to reach voters. One such is 36-year-old Siddhartha Chakravarthy, a Silicon Valley-trained engineer who has become a political entrepreneur and is in the fray in the Telangana state elections this month-end.

Voters can access his campaign by scanning a QR code by paying less than one rupee for access. It is a cost-effective campaign, but the question is: will such efforts click among people used to the dust and grime of a typical poll campaign? We will soon know.

While it is true that falsehoods have always been part of election campaigns since the dawn of elections, we have now entered a phase where sophisticated AI tools, not exactly humans, will manage and disseminate reams of disinformation. It is an uncharted turf.

Hence the anxiety that only gets aggravated by the forecasts of AI going rogue. Statements by technologists and cybersecurity investigators as well as politicians offer no scope for cheer.