$417 billion. That is the estimated global cost of disinformation in 2024, according to Sopra Steria, a figure comparable to the annual revenue of the world’s largest technology companies. In France, according to Marc-Antoine Brillant, head of Viginum, nearly 1,200 digital destabilization operations target the country every year. Faced with a threat that combines profit, political influence, and organized chaos, an ecosystem of countermeasures against information manipulation is taking shape. An overview of the initiatives gaining scale and what they cannot yet resolve alone.
Viginum, the DSA, and the European Democracy Shield: The Institutional Response
The institutional response is gaining momentum on both fronts (national and European) with growing coherence between the different mechanisms.
In France, Viginum, established in 2021 and attached to the General Secretariat for Defence and National Security, tracks foreign information interference and publishes public reports on identified operations, including Doppelgänger and Storm-1516. Since 2024, it has been collaborating with Arcom to enforce the DSA (Digital Services Act) against major platforms, including X and TikTok, which are currently under investigation for failing to meet their transparency obligations.
At the European level, the Commission took a further step in November 2025 by presenting the European Democracy Shield. This framework aims to go beyond the DSA by coordinating member states’ responses to information manipulation campaigns. Meta’s decision in early 2025 to shut down its fact-checking program, which funded more than 80 media outlets worldwide, paradoxically accelerated this consolidation.
Media Literacy and Pre-Bunking: The Most Robust Defence Against Disinformation
Behind the technical and institutional mechanisms, researchers agree on one point: the most durable defence against online disinformation remains media and information literacy. Not after-the-fact rebuttals, but what experts call pre-bunking, a technique of cognitive inoculation that exposes individuals to manipulation techniques and the cognitive biases exploited by disinformation campaigns, before they encounter them in the wild. The goal: to build lasting critical thinking reflexes and reduce the virality of false information.
The CLEMI (Centre for Media and Information Education) runs educational fact-checking workshops in middle and high schools across France. Through the De Facto programme, Sciences Po, AFP, and several partners have built an algorithm autopsy laboratory that allows students and citizens to directly test the impact of algorithmic biases on their daily information consumption. At Cambridge, the online game Go Viral! exposes users to the mechanics of viral disinformation to build immunity before any real-world encounter. The fight against disinformation targeting young audiences is today one of the most strategic priorities — and one of the best-documented areas of disinformation prevention research.
These approaches converge on the same conviction: you cannot verify everything that circulates. But you can build minds capable of doubting before sharing.
Fact-Checking in France: Initiatives That Detect Before the Crisis
For a long time, the response to disinformation was essentially curative: wait for a false piece of information to spread, verify it, publish a rebuttal. A useful model, but one structurally lagging behind the speed at which viral content propagates. The entire sector is now pivoting toward a preventive logic.
France’s fact-checking infrastructure rests first on solid, established players. AFP Factuel, founded in 2017, now deploys more than 140 journalists across 26 languages worldwide. A study conducted by Sciences Po and the University of Liège in partnership with AFP demonstrated that the rapid identification of a false piece of information reduces its propagation by 8% — a statistically significant result that, at the scale of billions of social media interactions, represents a massive brake on the spread of misinformation. In France, Vrai ou Fake (France Info) aggregates fact-checks from AFP, 20 Minutes, Libération, France Info, and Les Surligneurs on a single portal. Les Décodeurs (Le Monde) and Checknews (Libération) complete this editorial network.
In February 2025, Radio France united French public media around the Alliance for Facts, together with France Télévisions, France Médias Monde, TV5 Monde, and INA. This initiative merged in April 2025 with the Eurovision News Spotlight network of the EBU (European Broadcasting Union), which now brings together 18 European public broadcasters within a single collaborative verification infrastructure, the largest of its kind in Europe.
The most recent initiative marks a methodological turning point. Les Surligneurs and QuotaClimat have launched Droit à l’Info, an early detection system for misleading narratives, activated before they go viral. The goal: identify suspicious narratives on social media, online press, and audiovisual media, have them verified in real time, then distribute alerts through direct channels as soon as the propagation risk becomes significant. The system is being positioned ahead of the 2027 French presidential election. It is no longer “verify after”, it is “detect before.”
Generative AI: A Disinformation Threat and a Detection Tool
The picture would be incomplete without addressing the central contradiction of this moment: generative artificial intelligence simultaneously accelerates the production of false information and the tools to combat it.
On one side, large language models (LLMs) make it possible to generate at scale credible, misleading content (texts, images, voices, videos) linguistically and culturally adapted to each target. According to NewsGuard‘s monthly barometer, which tests the ten leading chatbots on sensitive topics (health, politics, immigration), the rate of false claims repeated by AI chatbots rose from 18% to 35% in one year, nearly double. These findings are now cited by European regulators and the US Congress in their work on regulating generative AI.
On the other side, AI for disinformation detection is becoming a frontline instrument in its own right. The consortium Science Feedback, QuotaClimat, and Data For Good has deployed Climate Safeguards, a tool that scans TV and radio broadcasts in real time to detect climate disinformation. Across the 18 channels monitored in the first quarter of 2025, it identified 128 verified cases, more than ten per week. Online media outlet Les Électrons Libres went a step further: on May 27, 2026, it launched Céleste, an editorial AI deployed on X, powered by Mistral and trained on the outlet’s editorial content. Its stated goal: to offer a francophone alternative that is transparent about its sources, where tools like Grok have accumulated controversy over disinformation. The technological race between AI and disinformation is underway. Its outcome remains open.
Limits of Anti-Disinformation Initiatives: What Remains to Be Solved
All of these responses (early detection, regulation of digital platforms, media literacy, algorithmic verification) share a common logic: they act on content that has already been published, or on the minds that receive it. None resolves the problem upstream of publication. This is the structural limitation of the entire current anti-disinformation ecosystem.
The question that remains unanswered is this: how do you certify the authenticity of content at the moment of publication, and prove it genuinely comes from who it claims to come from? A rebuttal, however swift and well-sourced, always arrives after the propagation. And faced with a convincing voice or visual deepfake, no fact-checking tool or verification platform can substitute for a digital proof of authenticity established before the attack.
This is precisely the barrier that real-time content certification seeks to lift: not detecting the fake after the fact, but proving authenticity before distribution, regardless of format, platform, or production tool. The countermeasures are industrializing. So is information manipulation. The gap between the two is now determined in the seconds separating the publication of a piece of content from its first viral spread.
Detecting the fake after the fact remains essential. But the next frontier of the fight against disinformation lies upstream, at the moment a piece of content is certified at the source, not at the moment it is contested.
Discover how the Certiphy.io solution fits into this ecosystem →
