Muthuvel Karunanidhi is the former leader of the Dravida Munnetra Kazhagam (DMK) party, a centre-left party currently in power in Tamil Nadu. It shouldn’t have been a surprise to see him at the launch of the book Karunanidhi: A Life In Letters. Afterall, the event was organised by the DMK, and the book was a collection of his writings, speeches and poems. That his son MK Stalin, current leader of the DMK, was there shouldn’t have raised any eyebrows either. But it did, because M Karunanidhi has been dead for six years! 

The video was a deepfake made by Muonium AI, the Indian based media tech firm that was co-hosting the book launch. The resurrection of deceased politicians, whether for nefarious political gain or click-bate marketing, gives us a glimpse of where we are heading. The influence of AI is spreading. Is the election process under siege? Is democracy as we know it is at risk? We’re about to find out. 

Within the next twelve months, more than two billion citizens from 50 countries will go to the ballot box. There are General or Presidential elections in the United States, The UK, Indonesia, El Salvador India, Mexico and Pakistan. The European Union will hold its parliamentary elections in June. Staffan Lindberg, the director of the Varieties of Democracy, calls 2024 a ‘make or break year for democracy worldwide.’ 

But this ‘make or break’ year has been building. It's been four years since the deepfake of House Speaker Nancy Pelosi appearing drunk; seven since lip syncing software was being used to make videos of Barack Obama. That software ‘uses artificial intelligence (AI) to match audio of a person speaking with realistic mouth shapes, which it then grafts on to an existing video’. There’s an instruction manual online.  

But this is 2024, not 2017. Things have moved on. I’d ask AI for a quip right here, but it would just tell us all to ‘fasten our seatbelts.’ Buckle up. From the grass roots, all the way to the White House, in a bid for electoral advantage, politicians, authoritarian regimes, state-funded hackers and YouTubers are harnessing AI's capabilities to streamline campaigns, change public opinion and reshape the narrative of democracy. 

Gulp! 

Let’s start… here, with Andrew Grey, an independent running for office in England. He’s young, and open to the possibilities of using AI to hi-jack power. In fact, he’s so enthused he’s using it to write his manifesto and canvas residents on local issues. Further up the UK political food chain, a deepfake of then Prime Minister Boris Johnson and leader of the opposition Jeremy Corbyn released (‘released’, it’s not a bloody blockbuster.) in 2019 shows the embattled leaders giving their public support for each other. I know I said this was 2024, but 2019 deepfakes were so cute and fluffy. 

Politicians attacking politicians isn’t new, but the technology now gives them more ammunition, more fronts of attack. In July, a pro-Ron DeSantis super PAC used a fake version of Donald Trump’s voice in a television ad attacking… Donald Trump. In the New Hampshire primary a deepfake of Joe Biden was used to discourage democrats from voting.

From Presidential film reviews, to manipulated images of the British Prime Minister pouring a bad pint, you may laugh it all off. Many deepfakes aren’t particularly deep, many look decidedly fake, but don’t be lulled into submission: deepfake video and audio is being wielded in much more dark and nefarious worlds. Venezuelan officials are using AI to create fake media platforms, spreading fake news with avatars, manipulating the media narrative, driving a gigantic wedge between an already polarised population. Days after this video of Gabon President Ali Bongo surfaced, the military attempted a fucking coup. 

Most of us can find a laugh (even though we perhaps shouldn’t) when Donald Trump posts manipulated audio of Elon Musk berating Hitler, the Devil and Dich Cheney to his Instagram, but it really isn’t a laughing fucking matter when innocent people are being murdered in Central Africa. 

What are governments and tech companies doing to overcome the challenges and dangers? How do we stop more coups and rigged elections? How do we prevent Instagram from becoming a real life Saturday Night Live? How can we ever trust what a politician says again? MidJourney CEO, David Holz, says the AI image creation company is considering banning the creation of political images. “I don’t know how much I care about political speech for the next year for our platform.” That takes care of the memes. What about the $2.6 billion that’s spent on online political advertising in the US each year?  

Google has updated its political content and advertising policy. In most countries only advertisers who have been verified can run ads. Yet the tech giant still has ‘different requirements for political and election advertising based on region.’ Probably different in Gabon. Can’t find the policy. Maybe it’s been held hostage by a virtual politician.   

What we call deepfakes, Google calls ‘synthetic content’. And when it comes to synthetic deepfake content, advertisers on Google must… ‘prominently disclose when their ads contain synthetic content that inauthentically depicts real or realistic-looking people or events. This disclosure must be clear and conspicuous, and must be placed in a location where it is likely to be noticed by users.’ 

Our furry fluffy friends at Facebook already ban the use of deepfakes, but they rely on third parties and humans for verification and fact-checking, which is like trying to use snowballs to freeze the sun.  

In the EU, the A.I. Act is being used to establish ‘transparency’, ‘accountability’ and ‘awareness’. They love a good key-word, it’s like they write their policies for ethical search engine optimisation. 300 pages of legal frameworks that might one day be applicable… somewhere… aren’t going to slow down the fake-train. What’s that? Breaking news… Open AI have just released Sora…Are people still trying to convince us filters and watermarks offer tools to discern between authentic and fabricated material? 

Eleven Labs, whose technology was used for the Biden ‘don’t vote’ deepfake, are working on ‘addressing important concerns related to the misuse of AI and synthetic speech.’ I don’t know about you, but to me it all feels so vague, so wishy-washy, so… weak. Like it just won’t work. And the parts that do work, will be far too late. Any horses bolting any stables long since melted down for glue.  

The World Economic Forum has proposed four countermeasures:

1. Technology

“Multiple technology-based detection systems already exist today. Using machine learning, neural networks and forensic analysis, these systems can analyse digital content for inconsistencies typically associated with deepfakes. Forensic methods that examine facial manipulation can be used to verify the authenticity of a piece of content. However, creating and maintaining automated detection tools performing inline and real-time analysis remains a challenge.”

The stumbling block there is the last sentence. 

1. Policy Efforts

“We need international and multi-stakeholder efforts that explore actionable, workable and implementable solutions to the global problem of deepfakes. In Europe, through the proposed AI Act, and in the US through the Executive Order on AI, governments are attempting to introduce a level of accountability and trust in the AI value system by signalling to other users the authenticity or otherwise of the content.”

You can’t even play Mario on X-Box. 

1. Public Awareness

“Starting from early education, individuals should be equipped with the skills to identify real from fabricated content, understand how deepfakes are distributed, and the psychological and social engineering tactics used by malicious actors.”

Good luck with that. I wish… I really wish this could be implemented. Maybe deepfakes are the push we need to truly overhaul the western education system. 

1. Zero-Trust Mindset

"In cybersecurity, the “zero-trust” approach means not trusting anything by default and instead verifying everything. When applied to humans consuming information online, it calls for a healthy dose of scepticism and constant verification."

As far as I can work out, that involves curiosity, a diminishing resource.

The World Economic Forum really isn't going to save us.

Which leaves us with: 

Web3 has a community driven solution to everything these days. Maybe blockchain is the only way to protect democracy from the virtual clammy hands of the deepfake. Ethereum co-founder Vitalik Buterin (Hey Vitalik, if you read this, what would it take for you to appear on my podcast?) suggests a really complicated - obviously - solution could be the use of old jokes, family anecdotes and communal memories as a way to defend ourselves from imitation and fraud. 

Part of his solution involves shared experiences. (With the hyper-personalisation of gaming and media, shared experiences might soon be a thing of the past. But that’s for another day.) He probably hasn’t ‘cracked the deep-fake code’ - knowing the name of your invisible friend when you were six isn’t going to stop a North Korean pawn taking two videos off YouTube, splicing them together and creating some kind of political event horizon from which no truth can ever escape. Or something. Might stop fraudsters stealing your social media persona though. Or stop things like this happening. 

Whether blockchain can be used in the raging inferno against deepfakes is beyond my pay remit. But if we do manage to create a fair and transparent (ish) election process; if Google can clear the path for an honest campaign advertising trail; if Open AI (lol), MidJourney and Dall-E can somehow block the creation of images and videos and text; if you can indicate whether audio has been voiced by a real person or a bot; If Muonium AI and Eleven Labs and their ilk can hold fire on the marketing; if politicians themselves can leave the goddamn prompts alone for five seconds, what about the day itself? What about the ballot box? Can we keep that safe? Can we guard the humanity of Ancient Rome in that? I mean, robots can’t vote yet. Deepfakes can’t close a curtain and ‘mark it X’. 

Sierra Leone held what is generally regarded as the first blockchain election. Or did it? “The National Electoral Commission (NEC) is the "sole authority" on Sierra Leone's public elections, and the group has gone out of its way to make it clear that it did not use blockchain in the March 7 election.” 

Which makes Thailand the first country to hold a blockchain election. Kind of. In fact, Thailand’s Democratic Party used Z-coin to elect its leader. “Live e-voting was held from November 1–9, 2018 and more than 120,000 registered Democrats voted via blockchain to elect former Thai Prime Minister Abhisit Vejjajiva by a margin of 67,505 votes.”

Which makes Switzerland the first country to hold a blockchain election. Or not. 72 people voted on made-up questions in a trial in the city of Zug. Oh Switzerland. Which makes Japan…Do you ever get the feeling that what you read and believe isn’t based on reality? 

Many will say the journey toward AI-resilient, deepfake averse democracies demands concerted efforts across technological, legal, and regulatory domains. When YouTubers can make videos of politicians saying anything they want in about eighteen seconds, those concerted efforts need stepping up. I don’t mean to be the bearer of bad news, but two billion people are going to the ballot box this year. Deepfakes and fake news coupled with a divided population easily manipulated by social media? Democracy might not be in such great shape in 2025. 

And since I’ve got this far, and nobody likes a pessimist, dare I say deepfakes are not the problem? To quote my friend Owen, ‘lobbying, self-centred policies based on capital gains, and the co-opting of the legal systems to protect the rich and powerful, has led to a situation where most people are fighting and working to survive and feed themselves and their families, and aiding industrialisation in the process, and causing the massive pollution that will kill the system.’ Most people really don’t care what politicians say, real or fake.