Full Fact & Fundación Maldita.es
Harmful misinformation is spreading faster than ever. Across Europe, false claims about health, politics and the environment are shared on short-form video platforms and social media with increasing speed. By the time a fact check is written and published, the damage is often already done.
That’s why Full Fact, Fundación Maldita.es and the European Fact-Checking Standards Network (EFCSN) have joined forces on a Prebunking at Scale project (PAS). The project is building a shared AI-powered tool and a common methodology for prebunking, rooted in the experience of more than 40 fact checking organisations across Europe.
The origins of PAS
Fact checkers have long relied on debunking - publishing corrections after a false or misleading claim spreads. Prebunking, by contrast, aims to prepare the public in advance by exposing them to the tactics and tropes of misinformation before they encounter it.
But until now, prebunking has faced two challenges:
Messages are often too broad to be useful in fast-moving debates
Prebunks are rarely published in the narrow window before narratives containing disinformation take off.
PAS was designed to solve both problems. By combining AI monitoring with human expertise, the project identifies early signals of emerging disinformation narratives, so fact checkers can publish prebunks in time to blunt their spread.
What the tool does
At the heart of PAS is a multilingual monitoring and analysis system. It monitors short-form video content from platforms like YouTube Shorts, TikTok and Instagram Reels and extracts claims using speech-to-text and text-on-screen recognition tools developed by Full Fact. These claims are clustered into narratives - emotive and persuasive storylines that cut across platforms and languages - using an AI extraction system developed by Fundación Maldita.es.
European fact checkers using the tool can:
Search and filter claims and narratives by date, topic, language, entity or virality.
Receive alerts when harmful and fast-growing narratives emerge in their region.
Track narratives over time to see how they evolve and spread.
Behind the scenes, the tool uses large language models and vector databases to encode, cluster and classify claims. But crucially, fact checkers remain in the loop. Their expertise validates outputs, ensures accuracy and contextualises findings.
A collaborative European effort
PAS is not just a piece of technology. It’s a collaborative infrastructure for the fact checking community. EFCSN and its members have developed a shared definition and methodology for prebunking, so organisations across Europe can align on best practices and build materials that are more consistent, timely and impactful. The multilingual monitoring and analysis system informs their editorial decisions and helps to allocate resources where they are needed most.
The first version of the tool supports five European languages. By next year, we will expand this to more than 20 languages.
Why this matters
Every false claim that goes unchallenged chips away at trust in democracy, public health and science. By shifting from a reactive to a proactive model of prebunking, we can give fact checkers and the public a fighting chance.
The work of tackling misinformation will always need human judgment, local expertise and cross-border collaboration. With Prebunking at Scale, we’re adding a new tool to that fight.
The Prebunking at Scale project is made possible with the support of the European Fact-Checking Standards Network.