Zum Inhalt springen

Disinformation and Misinformation

Zusammenfassung

The internet promised to democratize information. It also democratized falsehood. Misinformation (false content spread without intent to deceive) and disinformation (falsehood deployed deliberately, often as a weapon) are as old as communication itself — but networked platforms gave them unprecedented reach, speed, and precision. From Cold War “active measures” to the 2016 US election, from anti-vaccine movements to AI-generated deepfakes, the manipulation of the shared information environment has become one of the defining political problems of the digital age. This article traces how an old human problem was supercharged by social media, recommendation engines, and now generative AI — and why technical fixes keep colliding with the much harder questions of truth, trust, and free expression.

A Vocabulary for Falsehood

Researchers distinguish three overlapping categories, a taxonomy popularized by Claire Wardle and Hossein Derakhshan in a 2017 Council of Europe report:

  • Misinformation — false information shared without intent to harm (a person forwarding a wrong health tip in good faith).
  • Disinformation — false information created and spread deliberately to deceive, manipulate, or profit.
  • Malinformation — genuine information weaponized out of context (leaked private data, decontextualized real footage).

The crucial point is that the content’s truth value is only half the story; intent and context matter as much. This is why “fake news” — the phrase that exploded after 2016 — became analytically useless: it collapsed honest error, propaganda, satire, and partisan disagreement into a single epithet that politicians soon turned against journalism itself.

Old Tactics, New Infrastructure

Disinformation predates the internet by centuries. The Soviet KGB ran systematic “active measures” (активные мероприятия) throughout the Cold War — notably Operation INFEKTION in the 1980s, which planted the false claim that the US military had engineered HIV/AIDS, seeding it in a small Indian newspaper and laundering it through the global press until it appeared as established fact.

What changed was the infrastructure. Three features of networked platforms transformed the economics of deception:

  1. Zero marginal cost of reach. A single actor could address millions without a printing press or broadcast license.
  2. Algorithmic amplification. Engagement-optimizing feeds reward emotional, surprising, and outrage-inducing content — and falsehood, unconstrained by reality, is often more novel and emotionally charged than truth. A landmark 2018 Science study by Vosoughi, Roy, and Aral found that false news on Twitter spread “farther, faster, deeper, and more broadly” than true news, and that humans, not bots, drove most of the difference.
  3. Microtargeting. Behavioral data (surveillance capitalism) let propagandists tailor messages to psychological and demographic niches.

2016 and the Disinformation Panic

The watershed was 2016. Russia’s Internet Research Agency (a St. Petersburg “troll farm”) ran a coordinated influence operation across Facebook, Instagram, Twitter, and YouTube, reaching an estimated 126 million Americans on Facebook alone, according to the company’s testimony to Congress. The same year, Macedonian teenagers in the town of Veles ran hyper-partisan US political sites purely for ad revenue — disinformation as a business model, not ideology. The 2016 US election and the UK’s Brexit referendum turned “disinformation” from an academic term into a global political obsession.

A parallel front opened in public health. Anti-vaccine misinformation, long simmering since Andrew Wakefield’s retracted and fraudulent 1998 Lancet paper linking the MMR vaccine to autism, found an accelerant in social media. During the COVID-19 pandemic, the WHO declared an “infodemic” — an overabundance of information, true and false, that made it hard for people to find trustworthy guidance and that demonstrably cost lives.

The Generative AI Inflection

The next phase is synthetic media. Deepfakes — a term coined in 2017 from a Reddit user deploying face-swapping neural networks — made convincing fake video and audio cheap. Generative AI (large language models, image and voice synthesis) industrialized the production of plausible text and media at scale, threatening to overwhelm the signals humans use to judge authenticity.

A subtler danger is the “liar’s dividend” (Chesney and Citron, 2019): once people know any image or recording could be faked, bad actors can dismiss genuine evidence as fabricated. The deeper harm of pervasive synthetic media may not be that people believe false things, but that they stop believing anything — an epistemic exhaustion that corrodes the very idea of shared facts.

Dead End: The Centralized Truth Arbiter

The most tempting and most repeatedly failed response is the idea that a platform or government can simply be the arbiter of truth — a central authority that labels content true or false and removes the false.

Every version of this has run into the same walls. Fact-checking is slow, labor-intensive, and arrives after a falsehood has already traveled (the “bullshit asymmetry”: refuting nonsense takes far more effort than producing it). Automated detection is brittle and easily gamed, and it cannot adjudicate the genuinely contested or evolving claims — about a novel virus, an unfolding war — where the line between “misinformation” and “minority but legitimate view” is exactly what’s in dispute. Heavy-handed removal fuels accusations of censorship and increases distrust among the very audiences most prone to conspiracy thinking, sometimes lending banned claims a forbidden allure (the “Streisand effect”).

By 2025 several major platforms had retreated from professional fact-checking, with Meta announcing a shift toward a crowd-sourced “Community Notes” model (pioneered by Twitter/X) — an admission that top-down truth arbitration had become politically untenable and arguably counterproductive. The emerging consensus is that disinformation is not a bug to be patched but a systemic condition of cheap, algorithmically amplified communication — better addressed through transparency, friction, provenance standards (e.g., C2PA content credentials), media literacy, and institutional trust-building than through any centralized oracle of truth. There is no “off” switch for human credulity.

📚 Sources