Jump to content
Symbolfoto: Das AIT ist Österreichs größte außeruniversitäre Forschungseinrichtung

Research project for the detection of disinformation successfully completed with the participation of APA and ORF

AI-supported tool provides indications of manipulated digital content and targeted disinformation with just a few clicks - no prior technical knowledge required on the part of users

Credit: AIT/GettyImages

For two years, a consortium led by the AIT Austrian Institute of Technology researched an easy-to-use tool that supports users with the help of artificial intelligence in detecting disinformation on the Internet. This is because the enormous and constantly increasing flood of information on the Internet brings with it new dangers, such as disinformation campaigns, which in their abundance can no longer be detected exclusively manually. Now the Austrian research project "defalsif-AI", which started in October 2020 as part of the national security research program KIRAS, has been successfully completed. The solution developed was designed in an open way and can thus be easily further developed as well as integrated into other systems.

The steady increase of disinformation, potentiated by social media and other filter bubbles, and the development of effective strategies and tools against it is one of the central challenges of our time. The media companies APA - Austria Presse Agentur as well as ORF were involved as project partners alongside Enlite AI, Research Institute, the University for Continuing Education Krems and ministerial end-users from public administration.

The overarching project goal was not "finding the truth", but the protection of democracy and democratic institutions. "Artificial intelligence acts as a system that supports humans in checking and verifying digital content, but does not take decisions away from them," says Martin Boyer, project manager and senior research engineer at AIT Austrian Institute of Technology's Center for Digital Safety & Security.

Easy to use and transparent

To check or verify digital multimedia content (videos, photos, texts, etc.), the user uploads them to an online platform or enters the relevant web address, adds background information and presses the start button. Additional information such as "audio" or "portrait image" helps the system to select the right analysis methods and tools. The system then searches on a technical level for clues to disinformation or manipulated content. Afterwards, users are presented with the results in a simple and understandable way.

The tool currently consists of several modules that can be expanded or adapted in a modular fashion. "We have designed the tool to be future-proof, so that we can continuously add further analysis methods and components and in this way constantly develop the tool further," emphasizes Boyer.   The researchers are immediately following up on the results of the project. Five follow-up projects are already in the starting blocks, in which the focus will be on further automation and further development of the analysis methods, among other things.

Further information on the project: www.defalsifai.at
KIRAS security research program: www.kiras.at

defalsif-AI was funded by the Austrian security research programme KIRAS of the Federal Ministry of Finance (BMF).


APA - Austria Presse Agentur
Corporate Communications
Tel.: +43 1 360 60-5710

AIT Austrian Institute of Technology
Michael W. Mürling
Marketing and Communications
Center for Digital Safety & Security
Giefinggasse 4 | 1210 Vienna | Austria
T +43 50550-4126 | M +43 664 2351747 | F +43 50550-4150
michael.muerling@ait.ac.at | https://www.ait.ac.at