Primary voters in New Hampshire got a robocall from U.S. President Joe Biden asking them not to vote in the primaries. During the recent Pakistan elections, Imran Khan was in jail but his party still managed to circulate a video of him delivering a ‘victory speech’. The day before the 2024 Indian elections, millions of voters got a WhatsApp video message from Prime Minister Narendra Modi cautioning them against voting the next day due to a national security threat.
Imran Khan’s AI-generated victory speech. | Photo Credit: Youtube
Only two of the above three stories are true. No prizes for guessing which two. Artificial intelligence (AI) tools are being used — and misused — widely by candidates and political parties around the world. This year, with over 50% of the world scheduled to vote, the fear is that AI tools could undermine democracy globally.
AI tools can help peddle lies as the truth, and even the truth as a lie. Some have termed this ‘the liar’s dividend’. Liars, and those looking to confuse or just convolute things, could have a field day with today’s AI tools. No wonder that many world leaders, including Modi, have specifically identified deepfakes as a threat to democracy. (Deepfakes are videos created using AI tools that are hard to discern as fake.)
Amplifying the threat
How big and real is this threat? Is it being blown out of proportion? AI tools are not necessarily a new threat. Marketers and campaign managers have propagated fake videos and spread disinformation in the past as well. So what is different this time? AI is being seen not as a new threat, but a ‘threat amplifier’. It allows even small teams of individuals to ‘turbocharge’ their efforts to deceive voters by creating fake content much faster and in potentially a more targeted manner.
Author and podcaster Anirudh Suri | Photo Credit: Special arrangement
In all the hype about AI affecting elections in 2024 though, social media’s continued impact must not be put on the backburner. AI tools are used primarily to create the content, but social media platforms enable widespread distribution of the fake content. The ease of use of AI tools, combined with the distributive power of social media, could really accelerate election manipulation efforts at a scale that was unimaginable before.
Here is a simple example of how it might play out. The day before the election, you get a robocall, stating your name, polling booth and address of the polling booth you are supposed to vote at. Except, the polling booth name and location have been purposely mis-stated, so you end up at the wrong booth. You might even get an official-sounding WhatsApp message with the same fake information. A few thousand voters — likely to support an opponent — are specifically picked by an AI engine based on a sentiment analysis of their social media activity for this highly targeted exercise. Repeated in various Indian languages across different constituencies, it can change the election outcome.
The manipulation may also originate outside of India. Technology has become intertwined with geopolitics, as I have argued in my book The Great Tech Game. The geopolitical risks of AI-driven disruption to Indian elections are also very real. China, for one, has not shied away from leveraging technology to further its geopolitical objectives. The last few years of a frosty relationship with India might tempt China to fray the Indian electorate using AI-powered influence operations as well as deepfakes. Even if it is not capable of changing the outcome, an autocratic China would not want to miss the opportunity to create fissures and sow discord in a democratic India.
So is it inevitable that 2024 will be the year of ‘deepfake’ elections around the world, including India? No, not necessarily. Acknowledgement of the threat is the first step towards finding a solution. But many key players will need to come together to tackle this problem.
Preventing misuse
One, the Election Commission of India must implement strict guidelines for creation and distribution of such AI-enabled deepfakes in the run-up to the elections. Examples of such crackdowns exist around the world, including the U.S. Federal Communications Commission’s recent ruling of robocalls using AI-generated voices as illegal.
Two, Big Tech must play a proactive, constructive role in mitigating the threat. At the Munich Security Conference in February, several tech firms pledged to take reasonable precautions to prevent the misuse of AI tools in the upcoming elections. While this accord was seen by critics as insufficient and lacking in teeth, it was a good first step. Their election teams must work with their engineering and AI teams to develop guidelines and technical tools to prevent the widespread distribution of such content. Together, these firms must develop industry-wide standards for detection and labelling of AI-enabled content, with strict action being triggered when such content is political or electoral in nature.
External Affairs Minister S. Jaishankar in conversation with Atlantic Council President Fred Kempe on the sidelines of the Munich Security Conference 2024. | Photo Credit: ANI
If needed, the Election Commission should hold tech platforms, especially the social media platforms, accountable, including the possibility of potential penalties. With great power comes great responsibility. And, with deep pockets and large engineering teams comes great ability to find tech-driven solutions to such problems.
Three, governments must introduce necessary legislation to give teeth to such precautionary and punitive actions. Legislation might not be possible before the elections. But the problem isn’t going away, and therefore, legislation must be introduced sooner than later.
Four, political parties and candidates themselves must also sign up to voluntary pledges this year to not use deepfakes. Voters must also be made aware — through campaigns funded by the Election Commission or philanthropic efforts — of the dangers and prevalence of deepfakes.
A final point. The current national AI strategy of India highlights a key theme: #AIForGood. That is, India wants to use AI for the public good. There couldn’t be a better public good than for everyone to come together to find ways to use #AIForGoodElections.
Deepening democracy
Like any other powerful technology, AI can be leveraged for good or evil, depending on the intentions of the people using it. AI tools can deepen our democratic ethos and make our political discourse more representative. AI can be used to better identify issues that matter to different segments of voters. Politicians can use AI tools to communicate in languages beyond just the mainstream or official ones. AI can be used for fraud detection at the booth level. AI can be used for better mapping and location of election booths across the country, not to mention make them more accessible for the differently abled. The list is endless.
People naturally tend to focus on the dangers of new, emerging technologies. But the last decade has showcased India’s confidence with technology adoption, as evidenced by the successful penetration of smartphones and UPI. In 2024, the public discourse in India must encourage and incentivise key players, including tech start-ups and entrepreneurs, to leverage AI to make elections more democratic, not less. That would make the world’s largest democratic exercise a truly path-breaking one, and inspire democracies around the world.
The writer is a non-resident scholar at Carnegie India and host of The Great Tech Game podcast.
