Press release by EBU on the start of vera.ai.

EBU Takes Part In EU Project to Counter Disinformation with the Help of AI

Online disinformation and fake media content have emerged as serious threats to trust in media and ultimately society and democracy. And advances especially in AI-driven tools mean that in future such campaigns could become even more sophisticated, subtle and hard to detect. The vera.ai project, short for ‘VERification Assisted by Artificial Intelligence’ aims to level the playing field and help tackle this problem.

Kicked off last month and scheduled to run over 3 years, the vera.ai project gathers 14 experienced partner organizations including the EBU and Member organization Deutsche Welle, as well as research institutes, universities, commercial companies and the news agency AFP. Together, the consortium aims to enable the development of professional trustworthy AI solutions that can help to unmask and neutralize advanced disinformation techniques. The work will involve close collaboration between media professionals and researchers and is also intended to create a solid foundation for future research in the area.

The AI technology to be developed is meant to internalize concepts such as fairness, transparency and robustness against 'concept drift'. Another goal is to ensure that the tech can adapt to the evolution of disinformation, that it can be used a ‘fact checking tool in the loop’, and to make it work for a range of media types and languages. Naturally, tasks such as deepfake detection are part of the catalogue of required capabilities.

vera.ai is multidisciplinary and collaborative project, and will involve opensource algorithms. The mix of participating organizations is an advantage: For example, the project will benefit greatly from continuous access to ‘real life’ content collected through the InVID-WeVerify plugin and the Truly Media/EDMO platform, which will provide important data to train new AI models. And it can benefit greatly from the input provided and testing done by the professional journalists and fact checkers from project partners including DW, AFP and EUDL, the EDMO and InVID-WeVerify community members, as well as the members of associated media literacy, human rights and emergency response organizations.

The EBU participates in the project with its AI and Data Initiative (AIDI) group, the Eurovision News department, the Eurovision Social Newswire fact checking community, and the EBU Academy. One particular objective of the EBU is to ensure that new tools are fit for purpose in newsrooms.

“AI solutions are not meant to replace fact checkers. They’re meant to be tools in the fact checkers' arsenal against disinformation”, says AIDI coordinator Lalya Gaye. “And we need to make sure they integrate well into these professionals’ workflows. The EBU invites all its Members to take part in this project, whether it’s through surveys, design workshops, evaluations, hackathons, or training sessions.”

 

vera.ai is co-financed by the European Union, Horizon Europe programme, Grant Agreement No 101070093. Additional funding is provided by Innovate UK grant No 10039055, and the Swiss State Secretariat for Education, Research and Innovation (SERI) under contract No 22.00245.

 

You can find the original press release here.

vera.ai is co-funded by the European Commission under grant agreement ID 101070093, and the UK and Swiss authorities. This website reflects the views of the vera.ai consortium and respective contributors. The EU cannot be held responsible for any use which may be made of the information contained herein.