Online amplification networks detector
THE QUESTION
Can a user-friendly tool help detect hidden connections between disinformation websites and social media accounts at scale?
LOCATION: Bulgaria
SECTOR: Human rights and Democracy
TECH: AI
TIMELINE: November 2025 - Present
PIONEER: TBC
PARTNERS: Centre for the Study of Democracy
The Challenge
When CSD researchers discovered a network of 700 websites connected to a single company, they also discovered it to be Bulgaria's biggest disinformation spreader, with direct ties to Bulgarian media and politicians. It required weeks of painstaking manual investigation to map connections that should be detectable through automated systems. This desperately needed investigative work is out of reach for many journalists and fact-checkers across the region because they often lack the tools to do so efficiently.
These “mushroom” websites (coordinated, semi-automated networks of websites and social media accounts that generate, disseminate, and amplify disinformation) are increasingly shaping the information ecosystem in Bulgaria and the Western Balkans. Tracking tools exist but remain fragmented, expensive, and inaccessible to non-technical users that need them most: civil society organisations, policy analysts, and diplomatic actors.
The Idea
This pilot will develop an Online Amplification Networks (OAN) Detector—an integrated tool that simplifies complex technical processes into a user-friendly interface for non-technical practitioners. Building on CSD's decade of disinformation research and established relationships with European policymakers, diplomatic missions, and regional civil society networks, the detector will make sophisticated network analysis accessible to those on the frontlines of countering influence operations.
Key components include:
Website Network Mapping: Automated identification of hidden connections between websites through domain ownership analysis, sub-domain structures, and other forensic indicators currently requiring manual investigation
Named Entity Recognition: AI-powered identification of key actors, organisations, and narratives across multiple sources in Bulgarian and regional languages
Coordinated Activity Detection: Integration of existing methodologies to identify patterns suggesting centralised control or coordination across seemingly independent accounts and websites
Phased Platform Expansion: Starting with online media and website networks (CSD's strongest domain), then expanding to one social media platform, with eventual cross-platform capability
Non-Technical Interface: Designed specifically for European policymakers, foreign ministries, civil society researchers, and journalists without data science backgrounds
The pilot will test whether integrating existing analytical methods into an accessible tool can significantly accelerate the identification and mapping of coordinated disinformation networks, providing actionable intelligence to counter influence operations targeting European institutions and regional democracies.
Our learnings and stories so far
This pilot hasn’t started to publish yet, but there are plenty of other blogs to read below. Check back soon!

Gaps in WASH asset data mean that broken water points often go undetected, disproportionately affecting remote and marginalised communities. This pilot looked at how AI-assisted validation could strengthen oversight and ensure WASH infrastructure issues are addressed before they escalate.