In 1933, Joseph Goebbels stood at the Berlin Radio Show and declared radio ‘the Eighth Great Power’. He understood what we are only now beginning to grasp: new technologies don’t just change how we communicate; they redefine who can seize power, and how it is exercised. Radio offered a single, central channel through which a regime could define reality for an entire nation.

Goebbels moved fast. The Nazi state seized the Reich Broadcasting Corporation, imposed strict content rules, and subsidised Volksempfänger – ‘people’s receivers’ – at half the price of commercial radios so that every German household could tune in. Within months, millions of homes in remote villages heard the same voice from Berlin. Local newspapers and independent pamphlets became relics. The radio tower replaced the town square.

The innovation was less about content than about control. Radio collapsed Germany’s fragmented local media landscape into a centralised broadcast network where one voice spoke and millions listened. Coup plotters in the interwar period understood this instinctively: first seize the presidential palace, then the radio station (later the television studio). Control these two nodes, and you could master reality itself.

This pattern was not unique to the 1930s. Throughout history, political upheavals have followed revolutions in communication architectures. The advent of the printing press in the mid-15th century weakened the Catholic Church’s monopoly on religious truth, laying the groundwork for the Protestant Reformation. Radio and television allowed 20th-century governments – both democratic and authoritarian – to centralise the public sphere. When smartphones and social media reached mass adoption in the United States around 2012, establishment politics across Western democracies began to splinter.

Today, with generative artificial intelligence, we may be entering the most consequential communications upheaval yet: one that doesn’t merely redistribute power over information, but makes reality itself contestable.

Consider what happened when the broadcast ‘tower’ collapsed. In 2011, roughly a third of Americans owned smartphones. By 2013, that figure had surpassed 55 per cent. In a handful of months, the majority of citizens carried a device that could receive information, create content, and broadcast it to potentially millions – without editors, regulators, or institutional gatekeepers.

At the same time, platforms such as Facebook abandoned chronological feeds in favour of engagement-ranked algorithms. What users saw was no longer determined by recency or public relevance, but by opaque systems optimised for a single aim: keep them scrolling. Content that provoked anger, fear, or tribal outrage travelled fastest and farthest because the algorithms learned to treat those reactions as signals of relevance.

Political actors adapted to this new terrain with remarkable speed. Barack Obama’s 2008 campaign became a template for digital politics, harnessing platforms like Facebook, MySpace, YouTube, and Twitter not just to broadcast messages but to recruit volunteers, raise money, and turn supporters into online organisers. Bernie Sanders, a relatively unknown senator from Vermont, used emerging platforms such as Reddit and Snapchat to rise to national prominence under the hashtag #FeelTheBern.

By the mid-2010s, however, it was the populist right that appeared to have fully mastered the logic of algorithmic politics. In 2016, Donald Trump’s posts were shared five times more than Hillary Clinton’s. Brexit campaigners received twice the Facebook engagement of the Conservative Party. Germany’s far-right Alternative für Deutschland boasted twice the social-media following of Angela Merkel’s Christian Democrats, despite winning only half as many votes.

The systems themselves were not inherently right-wing; they were structurally biased toward norm-breaking, high-arousal speech. In an attention economy where visibility is power, those willing to provoke outrage or transgress convention enjoy a built-in advantage.

That advantage is now being consciously exploited beyond the populist right. In Britain, Green Party leader Zack Polanski has embraced TikTok-style videos and conversational podcasts to reach younger voters, helping to double party membership and positioning him as one of the few credible challengers to Reform UK’s Nigel Farage, whose TikTok following rivals that of all other Members of Parliament combined.

Across the Atlantic, New York’s new mayor, Zohran Mamdani, has turned viral video-making into an electoral art form, delivering messages on housing and affordability through TikTok and Instagram reels scored to laid-back hip-hop beats. The same engagement-optimising machinery that once supercharged the populist right is now being deliberately – and effectively – repurposed by the left.

History doesn’t repeat, but it rhymes. New communication architectures take a few years to permeate daily life before they transform politics. We are now four years into the era of mass-market generative AI.

When ChatGPT launched in November 2022, it reached 100 million users within two months – faster than any technology in history. But speed is not the crux of the story. Social-media changes the information people see by filtering and ranking existing content. Generative AI doesn’t simply distort information flows; by fabricating content that looks and sounds authentic, it renders truth itself into a triviality.

And yet, the ‘deepfake apocalypse’ has not quite unfolded in the way many experts predicted. Studies of recent elections suggest that cheap, low-tech manipulations – misleading captions, clipped videos – still outnumber sophisticated AI fakes. The immediate danger is not universal deception but pervasive doubt: a world where no proof is trusted, because anything could be fabricated.

Hannah Arendt once warned that the ideal subject of totalitarian rule was not the convinced believer, but the person for whom ‘the distinction between fact and fiction’ and between ‘true and false’ no longer exists. The goal of totalitarian propaganda was not simply to make people accept lies, but to make them stop believing in the possibility of truth at all. In her time, that level of confusion required a single voice speaking relentlessly from the centre of power. In ours, it can emerge from everyone, everywhere, all at once, with each piece of ‘evidence’ equally plausible – or equally suspect.

Confusion has become one of the blunt instruments of contemporary politics. When Steve Bannon described his media approach as a plan to ‘flood the zone with shit’, many treated it as the signature excess of one Trump‑era strategist. Today, that logic is increasingly pervasive across the spectrum: the aim is less to persuade than to paralyse, to saturate the public sphere with so much noise that holding on to any conviction becomes exhausting.

Decades ago, the media critic Neil Postman argued that Aldous Huxley, not George Orwell, had predicted the future. We would not be repressed by a Big Brother banning books, but lulled into passivity by a flood of trivial entertainment, until facts ceased to matter. The truth would not be censored so much as drowned, slowly, in complacency.

The central challenge is not simply to police ‘harmful content’ after the fact, but to rebuild what might be called the factual commons. Goebbels understood that controlling the radio network meant, for a time, mastery over Germany’s collective reality. Our predicament is perilous in a different respect. In a world where anyone can generate almost any version of events, the question is whether collective perception itself can survive.

The printing press shattered religious certainty, radio centralised political authority, and social media fragmented it. AI is doing something more radical, rendering truth into a malleable, personalised commodity. Each communications revolution rewrites the political order it inherits. Democracies will not endure by nostalgia for broadcast certainties, nor by chasing every viral falsehood. They will survive only by constructing the infrastructure through which facts can again be collectively known. We are watching the architecture of reality itself being rebuilt, brick by digital brick. The question is whether we can still agree on what counts as a foundation before the zone floods completely.