Digital democracy International

Synthetic Media Exposed: DRI’s Comprehensive Guide to AI Disinformation Detection

The release of ChatGPT, Midjourney, and other generative AI technologies have captured global attention and rocketed the debate around the future of artificial intelligence into the mainstream. However, as AI's capacity to generate realistic images, videos and audio increases, so too do concerns about the risks of these technologies being misused by malicious actors for the spreading of more convincing, sophisticated disinformation.   

In response to these risks, DRI has developed Synthetic Media Exposed: A Comprehensive Guide to AI Disinformation Detection, our very own compass for navigating the intricate field of advanced disinformation detection. This guide is part of the Disinfo Radar project, funded by the German Federal Foreign Office, and is designed to equip a broad spectrum of researchers, policymakers, and members of civil society with the tools they need to identify synthetic disinformation online. In it, readers will learn how to better recognise AI-generated content using manual methods such as glitch detection, as well as become familiar with more sophisticated tactics such as metadata analysis and AI-powered detection tools. Find the full guide below: 

Read the full guide

Guide Contents 

  • A glossary through which readers can become familiar with the more technical terms used in the field of disinformation detection. 
  • An introduction to the manual approaches for identifying AI-generated content, with a specific focus on glitch analysis. Here, we use concrete video, image, and audio examples to demonstrate which visual or auditory markers often reveal the presence of generative AI. 
  • A guide to how researchers can discern clues about generative AI usage by exploring the metadata within files. 
  • An explanation of the more sophisticated and contemporary approaches to AI detection, such as the use of machine learning and content authentication methods.  
  • Links to additional resources where users can learn about detection methods and initiatives.  

A Valuable First Step in Detecting AI Disinformation 

As AI-generated content floods the Internet, the increasingly AI-saturated media landscape demands a greater number of committed, knowledgeable fact-checkers ready and able to identify artificially generated disinformation. DRI’s guide is intended to address this need by functioning as an introduction for prospective disinformation researchers.  By outlining the most modern practices for synthetic content detection and defining the field’s more technical terms, it is hoped that this guide will de-mystify detection methods and equip readers with the know-how to identify disinformation on their own.  

Launch Event 

To celebrate its launch, DRI organised a webinar where the newly published guide was presented and reviewed by Bellingcat's Aiganysh Aidarbekova. The review was followed by a panel discussion with experts and advocates on the field of disinformation and AI governance, as well as a Q&A session with the online attendees. You can watch the recording below.



- Our guide was presented and reviewed by Aiganysh Aidarbekova | Bellingcat Researcher and expert -  


- Chris Starke | University of Amsterdam, Assistant Professor "The Human Factor in New Technologies"  
- Lena-Maria Böswald | Das NETTZ, Advocacy Manager Online Hate Speech and Disinformation  
- Athandiwe Saba | Code for Africa, Managing Editor  
- Luca Nannini | CiTIUS, University of Santiago de Compostela, PhD student in Explainable AI & Immanence, Researcher  

Democracy Reporting International's Disinfo Radar project, funded by the German Federal Foreign Office, aims to identify and address disinformation trends and technologies. Kindly take a moment to share your insights by participating in our survey. You may also register for our newsletter (select Digital Democracy to receive our Digital Drop, a newsletter dedicated to this topic). Your feedback contributes to our ongoing efforts in enhancing our research and promoting informed discourse.


20231013_Synthetic-Media-Exposed Download

This work is supported by