vera.ai has come to an end – its results remain accessible. More right here!
After three years and two months, the vera.ai project has formally come to an end on 31 October 2025. Since September 2022, 14 European partners have passionately worked towards the project's goal of building trustworthy AI solutions against advanced disinformation techniques, co-created with and for media professionals, and to set foundations for future research in the domain of AI against disinformation and related topics.
Although vera.ai is formally over, a lot of its work, learnings, outcomes and results will live on beyond the project's runtime - in various ways and form (for more see also below at the end of this article).
Among the project results are a variety of tools and services. In this article, we collect and provide further information about some of them - another legacy of vera.ai. Unless noted differently, these will remain accessible to the public in the foreseeable future.
vera.ai tool factsheets
The following table contains factsheets for tools and services that have been developed and/or have been improved during the project's life-cycle*. Factsheets contain information on (1) aim of the tool; (2) access options; (3) contact information; and (4) more information.
| Tool / Factsheet | Short description | Responsible partner |
|---|---|---|
| Audio Provenance Analysis | The Audio Provenance Analysis tool enables media professionals and journalists to verify the authenticity and reuse patterns of audio materials. | IDMT |
| B-Free | B-Free is a synthetic image detector based on a novel bias-free training paradigm | UNINA |
| Central Claim Extraction & Narrative Detection | The tool automatically extracts check-worthy claims and visualizes their narrative hierarchy. | KInIT |
| Chatbot-Assisted AI Explainability and Feedback Collection | A chatbot that helps media professionals understand underlying AI models. | USFD |
| Coordinated Sharing Detection Service | The Coordinated Sharing Detection Service (CSDS) helps media professionals, analysts, and researchers uncover networks of accounts that amplify identical or near-identical content across social media. | UNIURB |
| CooRTweet | CooRTweet is an open source R package for detecting and analysing coordinated behaviour on social media. | UNIURB |
| Database of Known Fakes (DBKF) | The DBKF allows to easily double-check whether a claim has already been debunked by trusted fact-checkers | Ontotext / Graphwise |
| Image Forgery and Localization Service | This service integrates multiple AI-based models to automatically detect and localize image forgeries. | CERTH |
| Image-Text Decontextualization Approach | This approach aims to label information that has been divorced from its original context, leading to distorted or misleading interpretations | CERTH |
| Keyframe Selection and Enhancement Service (KSE) | The KSE automatically extracts keyframes and enhances important visual elements – such as faces and text – to make details clearer for inspection. | CERTH |
| Location Estimation Using Solely Visual Clues | This service aims to infer the depicted location of an image – using solely its visual content | CERTH |
| Multilingual News Article Framing Classifier | This service identifies the main frames used in a particular article. | USFD |
| Multilingual News Article Genre Classifier | This service is able to identify news genres in 104 languages. | USFD |
| Multilingual News Article Persuasion Technique Classifier | This service is able to identify the main persuasion techniques used in a particular article. | USFD |
| Near Duplicate Detection Service | This service is a robust and versatile audio/visual search solution offering multiple search modalities. | CERTH |
| POI-Forensics | An audio-visual person-of-interest deepfake detector | UNINA |
| Syntetic Image Detection | This service integrates multiple AI-based models to automatically detect and analyze AI-generated or manipulated images. | CERTH |
| Synthetic Speech Detection | The Speech Synthesis Detection tool helps media professionals and journalists accurately identify AI-generated speech. | IDMT |
| Topical and Temporal Analysis Tool | The topical and temporal analysis demo is designed to assist media professionals, journalists and social media analysts in exploring how disinformation narratives develop and interact over time. | USFD |
| TruFor | A forensic tool for image manipulation detection. | UNINA |
| Truly Media | Truly Media is a collaborative verification platform that enables teams to work together in real time in order to collect, organize, and verify digital content. | ATC/DW |
| Verification Plugin | The Verficiation Plugin brings together tools (such as a set of tools developed within vera.ai) that support the verification of digital content. | AFP |
| Video Deepfake Detection Service | This service integrates two complementary deepfake detection models. | CERTH |
*list is not exhaustive
More results
For more results and outcomes, visit our publication section, check out the presentations that were held and are publicly shared, or have a look at the code and datasets we made available. Furthermore, we invite you to read through the articles written about various activities throughout the project duration.
Author: Anna Schild (DW)
Editor: Jochen Spangenberg (DW)
Factsheets: Provided by responsible partners
