Written by EFA intern Sara Doczi.
Every scroll is a vote of sorts – a like, a share, a moment of outrage – and most of us have no idea who’s counting. This International Day of Democracy falls on 15 September 2026. Perhaps today you might think about how online spaces can change how you participate in politics.
International Day of Democracy
Since 2008, the United Nation’s International Day of Democracy provides an annual opportunity to reflect upon the state of democracy around the world. But it now seems this reflection must shift. Understanding democracy today means understanding digital technologies and online spaces, and how their relentless advancement is reshaping our political institutions, often in ways we don’t fully see and not always for the common good.
Examining democracy gets exponentially harder once the online world enters the picture. Instead of looking only at the behaviour of states, we’re faced with millions of individuals, all operating on platforms built and owned by private companies. Real-life political events are now largely shaped by discourse on social media. So how accountable are the users posting it, and the platforms hosting it, for what follows?
The post-fact era
Social media discourse is fundamentally different from the discourse of legacy media. Throughout the past twenty years, customised and curated digital content set off a wave of hoaxes, fake news, and conspiracy theories that would eventually crest into a today’s landscape of mis- and disinformation, where shared, verifiable information stopped being the default condition of public discourse.
The nature of truth has fundamentally changed in an epistemological sense: in the post-fact climate, discourse starts with questioning the very existence and validity of the information we receive.
Politics and democracy will never truly be the same as they were in the era before true information became a privilege, rather than a given; an era when the term “fake news” could not be used as a valid political response to a claim one disagreed with.
How algorithms sort discourse
Part of the problem is that political discourse online feels like a mirror of the public, when it isn’t one. Algorithms work in mysterious ways: two people’s feeds in the same household can look nothing alike, shaped by invisible sorting neither of them controls. Local politics gets discussed on a global stage, with no reliable way to tell who else is standing in your corner of the algorithm, or how many of them there are.
That already shaky foundation runs straight into a second problem: monetisation. Algorithms reward what people engage with – not just by giving it a wider audience, but often by paying its creators for the privilege. The incentive is clear: post what makes people angry, and the anger pays.
Anger, clicks, money
“Rage-baiting,” the act of deliberately posting content to elicit outrage as a way to increase traffic and revenue on one’s social media page, is an increasingly widespread manipulation tactic. Often, this looks like creators playing obnoxious characters on social media. One creator reportedly made around $150,000 (roughly A$242,520) in 2024, largely through internet commenters telling her she needed to “bring herself down.”
Political discourse is a gold mine for accounts making outrage-inciting claims. In the lead-up to the 2024 US election, political accounts on X (formerly Twitter) could reportedly make “in the low thousands” of dollars by posting provocative and controversial material, such as content depicting former US President Joe Biden as a dictator.
The tactic of ‘engagement farming‘ travels well beyond one country or one issue: a now-suspended X account drew millions of views from racist posts aimed largely at Indians, while a handful of monetized accounts spreading misinformation after the 2024 Southport attack in the UK racked up 260 million views combined.
X-celerating monetisation
X, under Elon Musk, is particularly well-built for this kind of profit. Musk’s own record adds to the picture. In the final weeks of the 2024 race, his America PAC offered $1 million daily payouts to registered swing-state voters who signed a petition supporting free speech and gun rights, a scheme that drew a Justice Department warning about vote-buying laws and a since-dismissed lawsuit from the Philadelphia District Attorney. No charges were ultimately filed, but the episode underscored how loosely political money and platform incentives can mix under Musk’s ownership. Most platforms have rules against misinformation that can trigger demonetisation; X largely doesn’t enforce the same restrictions.
That gap leaves generative AI almost entirely unchecked on the platform. While Grok’s (X’s built-in AI) image generation has recently been limited, and the platform labels AI-generated content in some instances, AI remains a favoured tool for anyone chasing high engagement.
Uses of AI: fabrication and fine-tuning
Most obviously, AI can help users create entirely false material, such as the spread of a doctored image depicting Kamala Harris as a McDonald’s employee. Even Donald Trump used this tactic ahead of the 2024 presidential race, posting an AI-generated image suggesting Taylor Swift had endorsed him. Swift later explicitly endorsed his opponent, Kamala Harris.
However, AI’s more dangerous use may be subtler than fabrication. A London School of Economics and Political Science study found that racist and conspiratorial posts inciting riots after the 2024 Southport attack often used generative AI to “increase the viral potential” of their content. As social media users scroll through their feeds, AI assistance can make a real difference in whether a post stands out and garners an instant reaction…especially when the content is already controversial.
Local issues, global perspectives
The spread of engagement-farming also presents a newer threat: making money from disruptive content on social media is a tactic that can be deployed from anywhere in the world. Since X introduced a new transparency feature in November 2025, users have been able to see which country or region different accounts are based in.
People following pro-Trump and MAGA content discovered that many of the biggest accounts weren’t American at all, some traced to Thailand, Nigeria, and Bangladesh. One account, “MAGA NATION,” with nearly 400,000 followers and multiple daily posts, turned out to be run from Eastern Europe.
It’s not hard to see how this misleads American voters. In a BBC article, Alexios Mantzarlis of Cornell Tech agrees that making money is often a motivation for these accounts. People looking for genuine political conversation may believe they’ve found it on X, when in fact many of the loudest voices, foreign or domestic, are there to profit from the noise, not to participate in the debate.
Looking forward: navigating social media through recognising curated outrage
In a world where political conversation increasingly plays out on social media, completely retreating from online political life may not be helpful, and it isn’t realistic either. Social media is a key entry point to politics for many, and even users who are there purely to watch videos or keep up with friends inevitably encounter this content. The algorithm doesn’t sort for intent, only engagement and monetization.
But it does mean approaching that conversation with a clearer eye. Knowing that a feed is curated rather than representative, that outrage is often the product rather than the byproduct, and that a popular account may have no real stake in the country it claims to speak for, changes how a post should be read.
Social media regulation remains at a major crossroads: how far can we regulate content posted online without violating the fundamental (but universally variable) principle of free speech?
Censorship on social media faces not only the battle of free speech against recognising harmful posts, but perhaps more importantly, the non-democratic nature of social media platforms as private ventures. X, Meta, Google, are all companies with leaders that weren’t elected or chosen by the population in any way, and whose interests can largely be summed up as “increasing profits”.
It seems as though both regulation and non-regulation face the same problem: being executed in a way that is intended to benefit the platform, and not the user. This, users cannot change. Instead, we can change our participation, and in certain instances, our gullibility. As long as there is participation in a platform’s lucrative mechanisms, those mechanisms will prevail.
Democracy has always depended on citizens being able to reason together. The challenge of this moment is learning to do that in spaces built, at least in part, to profit from the opposite. Not falling into the trap of mindless, non-critical thinking based engagement and monetisation may be the first step in creating online spaces truly fit for democratic discourse.
Image credit: Mariia Berezovsky