What 392 Fact-Checks Tell Us About This Fortnight's Misinformation

30 fact-checking organizations across 6 continents. One automated intelligence report. Here's what they found.

Every week, thousands of false claims ripple across social media. Fake photos of politicians with Epstein. AI-generated videos of world leaders saying things they never said. Fabricated screenshots designed to trigger panic. Old photos recycled with new captions to stoke outrage.

But who's actually tracking all of this? And where?

I built an automated intelligence report that pulls in every fact-checked claim from the past two weeks — not just the image fakes, but all misinformation: false text claims, misleading statistics, conspiracy theories, out-of-context media, fabricated quotes, the lot. Then it categorizes each one using AI.

What the data shows

The biggest category? Fabricated text claims (96 cases) — plain false stories with no media manipulation at all. Second: out-of-context media (66) — real photos and videos repackaged with false narratives. AI-generated images came third (53), followed by old media passed off as new (45).

The dominant intent across all 392 claims: political manipulation (198 cases, more than half). South Asia was the most targeted region (166 claims), followed by North America (85) and claims with global reach (57).

24 claims scored the maximum severity of 5 out of 5 — including AI-generated videos of a world leader allegedly promoting a different religion, fabricated Epstein connection photos targeting politicians, and fake military attack footage designed to inflame tensions between nations.

How it works

The report is built on the Google Fact Check Tools API, which indexes structured claim reviews from organizations certified by the International Fact-Checking Network (IFCN). These are not random blogs — they are professional verification outlets that follow transparent methodologies.

The system queries the API with broad search terms to capture every available fact-check from the reporting period, deduplicates the results, and sends them to Gemini 2.0 Flash for multi-dimensional classification: claim type, generation method, subject category, likely intent, geographic target, platform of origin, debunking method, severity, sophistication, and emotional targeting vector.

The 30 fact-checking organizations

Here are the sources that contributed to the January 26 – February 9, 2026 report, ranked by number of claims reviewed:

All of these organizations are signatories of the IFCN Code of Principles or equivalent standards, and publish machine-readable ClaimReview markup that the Google Fact Check API can index.

See the full report

The report includes severity distribution, statistical breakdowns by claim type, generation method, subject category, intent, geographic target, platform, debunking method, sophistication level, and emotional targeting — plus the full list of all 392 claims with links to the original articles.

Built by Henk van Ess as part of ImageWhisperer.
Data sourced from the Google Fact Check Tools API. Classification by Gemini 2.0 Flash.