Verification Plugin
One of the assets of the vera.ai project is to be able to build new AI-tools, and enhance existing AI-features integrated in the Fake News Debunker toolkit (or verification plug-in) developed in previous EU-funded projects, namely InVID (2016-2018) and WeVerify (2018-2021).
No less than 20 tools for video and image verification, search and data analysis are presently supporting the work of more than 80 thousand active users per week (109 thousand active users per month) at the time of writing (mid-November 2022).
The toolkit has attracted a large community of users such as fact-checkers and journalists, researchers, academics, media scholars teaching media literacy, OSINT investigators, law and enforcement agencies, NGOs, etc…).
Designed as a kind of multi-purpose “Swiss army knife for verification” by fact-checkers for fact-checkers, in 2021 the browser extension won the 1st prize of the US Paris Tech Challenge awarded by the Global Engagement Centre of the US State Department and the Digital Forensic Lab of the Atlantic Council. The team behind the services came first in a worldwide competition in which about 40 teams participated.
The “Fake News Debunker” (aka verification plug-in or browser extension), coined by the Poynter Institute as “one of the most powerful tools for spotting misinformation online” has also been widely used during the highly disputed 2020 US presidential election to debunk disinformation, and more recently during the Russian invasion of Ukraine and many other events the world over.
Several of the verification plug-in features like keyframe fragmentation or OCR (Object Character Recognition) are already based on AI technology. They will be further enhanced in the course of the vera.ai project work.
Let us take a closer look:
The keyframe fragmentation service allows you to fragment a video into meaningful keyframes for each sub-shot to facilitate image similarity search and retrieval of information. This shows you whether the same or a similar video has been previously indexed by some of the seven reverse image search engines that are incorporated into the tool.
The OCR service, in turn, allows for retrieving text within images, thereby helping users to make sense of images spreading on social networks. This can be used to either locate events or to understand (through automated translation) what the event is about.
The vera.ai project consortium also plans to add more new AI-based tools to the toolkit, such as a deepfake detection service, audio forensics, improved image forensics, better archiving of disinformation as well as disinformation detection and enhanced social network analysis, and an improved Database of Known Fakes (DBKF).
Launched initially in 2017, the Fake News Debunker or verification plug-in has been designed and is being maintained by AFP Medialab since its launch, with the active support of all project partners past and present, namely scientific as well as media partners who are developing and providing innovative services and play an active role in co-development e.g. through requirements assessment and user testing or software evaluations.
*****
The plug-in works with all Chrome-based browsers (Chrome, Chromium, Edge, Opera, Brave) as a so-called browser extension. To get your version, download it from the Chrome store and start verifying right away.
*****
Author: Denis Teyssou (AFP)
Editor: Jochen Spangenberg (DW)
