Scientists from TUD Dresden University of Technology, HAW Hamburg, Alpha Strike Labs GmbH and Traversals Analytics and Intelligence GmbH are combining their research prowess to protect the internet from AI attacks as part of their joint BMBF project AI.Auto-Immune, which is receiving EUR 3.79M in funding. TUD is coordinating the project and receiving EUR 1.1M in funding.

Artificial intelligence (AI) is ideal for collecting and analyzing huge amounts of data, and providing a basis to then act on those insights. Attackers exploit this process to threaten IT systems. AI is therefore increasingly becoming an important focus of cyber security.

Attackers can use highly developed AI not only to search for specific and known vulnerable areas, but also to automatically explore the porosity of IT infrastructures. The AI employs a variety of methods to attack these infrastructures and develops ways to circumvent automated protection mechanisms based on known signatures or patterns.

The joint project AI.Auto-Immune aims to protect the internet and its services from such AI-based attacks. The team of researchers will work on new methods to evaluate communication data and network flows on the internet using AI algorithms to help uncover any weaknesses. An intelligent analysis of network services helps to identify patterns and emerging new threats at an early stage, so that security measures can be adapted accordingly. For example, the AI can automatically predict weak points and suggest recommendations for protection by generating images of the vulnerabilities of certain network areas, such as internal company networks or critical infrastructures.

Matthias Wählisch, Network Coordinator and Professor of Distributed and Networked Systems at TUD, is delighted about the project’s approval: "Strengthening the security of globally distributed and networked systems is hugely important and will be a task that accompanies us in the long term. We can only win the race against malicious hackers if we understand the limits of AI-based attacks on the one hand, and at the same time use AI to react flexibly and swiftly to emerging threats on the other." His team, alongside scientists from the Chair of Machine Learning for Computer Vision headed by Björn Andres, will focus on internet measurements and clustering methods for securing internet services. In particular, the aim is to recognize conspicuous patterns in communication services, even if they have not occurred before.

"AI.Auto-Immune is not only crucial for continued scientific success against AI-based attacks on the internet. The project is also of the utmost importance for our university as an institution and for Dresden as a science hub," emphasizes Rector Ursular Staudinger, extending her thanks to the participating researchers at TUD for their extraordinary commitment in this field.

AI.Auto-Immune is funded by the German government's research framework program on IT security, "Digital. Secure. Sovereign.” (German: Digital. Sicher. Souverän) as part of the measure "Secure future technologies in a hyper-connected world: Artificial intelligence” (German: Sichere Zukunftstechnologien in einer hypervernetzten Welt: Künstliche Intelligenz).

Contact:

Prof. Matthias Wählisch
Tel.: +49 351 463-38261
E-Mail