EU Funded Projects On AI In Dialogue With EDMO

The EDMO community and its Europe-wide network of hubs have started cooperation with six European Commission co-funded research projects, which include research on AI methods for countering online disinformation.

In more detail, the focus of ongoing research is on detection of AI-generated content and development of AI-powered tools and technologies that support verification professionals and citizens with content analysis and verification.

vera.ai

The project brings together research capabilities from different fields, fact-checkers, NGOs, and the European Broadcasting Union (EBU). It is developing tools for detecting AI-generated content and employs novel AI and network science-based methods to assist verification professionals. The new AI techniques developed in vera.ai will address key user needs of fact-checkers, including deepfake detection and transparency and explanation of AI outputs. First results include AI methods and tools for:

detecting local manipulations in synthetic images;

analysing traces left by generative AI on created images;

deepfake detection of persons of interest;

generation and detection of synthetic text using language models.

AI4TRUST

The project aims to enhance the work of human fact-checkers through automated platform monitoring of social and news media using advanced AI-based technologies. The project is building on proven tools and technologies, including an observatory that gathered COVID-19 disinformation, a clickbait content analyser and a “Verdict Generator” that could aid in debunking disinformation.

TITAN

The project aims to create an AI-based citizen coaching ecosystem against disinformation. It will be delivered through an intelligent chatbot that guides citizens in probing the factual correctness and reliability of online content. The chatbot would consider the citizen’s critical thinking capacity, incorporate fact-checking processes and tools, and emphasise a human-centred approach.

AI4MEDIA

This is 4-year project, which aims to establish a Centre of Excellence on AI research and engage with a wide network of researchers across Europe and beyond. It focuses on delivering the next generation of core AI advances and training to serve the media sector. At the same time it puts focus on ensuring that the European values of ethical and trustworthy AI are embedded in future AI deployments.

AI-CODE

The project is developing generative-AI-based solutions dedicated to media professionals. It reflects on rapid technological advancement not only in the area of generative AI but also in next-generation social media, e.g. AI-based decentralized and immersive virtual environments (like fediverses and metaverses). The project aims to provide media professionals with novel AI-based services to coach them how to work in emerging digital environments and how to utilize generative AI effectively and credibly.

AI4DEBUNK

The project will develop 4 human-centred AI-powered interfaces: a web plug-in, a collaborative platform, a smartphone app, and an AR interface. All these solutions will be built upon a “debunking” API, focused on 2 particular disinformation use cases: the war in Ukraine and climate change. The web plugin will notify citizens when they view already debunked content; the collaborative Disinfopedia platform will allow citizens to report questionable content; the phone app will support disinformation verification on mobiles; the AR interface will support citizens in the next generation social media environments.

Joint Initiatives and Available AI-Powered Tools

Researchers from these projects have joined forces and just published a new white paper: Generative AI and Disinformation: Recent Advances, Challenges, and Opportunities.

Verification professionals and disinformation researchers from the EDMO community and beyond can already benefit from some AI-powered image, video, and text verification tools released by these projects:

The free InVID-WeVerify content verification plugin, now integrates AI-powered verification tools developed by partners in the vera.ai project.

EDMO’s TrulyMedia platform is currently integrating the latest AI-based tools from vera.ai as well.

There are also opportunities to engage with these projects through their annual joint event. The inaugural event “Meet the Future of AI: Countering Sophisticated and Advanced Disinformation” was held in June 2023 in Brussels, where researchers from four of these projects (vera.ai, AI4Trust, TITAN and AI4Media) discussed the numerous barriers and research challenges in AI and disinformation, as well as agreed on a number of joint follow up actions. A summary of the conference discussions and main outcomes is available here.