Using AI to Tackle Disinformation: Methods and Tools from the vera.ai Project

As generative AI becomes increasingly fluent and accessible, its misuse for creating convincing and cost-effective disinformation campaigns has also been growing. From deepfake videos targeting political figures to AI-generated images and audio clips spreading false narratives, recent examples highlight the urgent need for robust detection mechanisms. This chapter explores the development and application of AI methods for detecting synthetically generated disinformation across various modalities - including images, video, audio, and text - within the vera.ai project. It begins with an overview of these methods, followed by a presentation of the AI tools that have already been integrated into the widely used open-source Verification Plugin, used by over 140,000 verification professionals worldwide. The chapter then introduces Truly Media, a platform for collaborative verification that is also incorporating vera.ai’s AI-based models. It concludes by highlighting the remaining challenges in this field and emphasizing the need for continued funding and collaboration to advance the development of models, tools, and datasets for AI-driven disinformation detection and analysis.