The project centres a collaborative approach in which experts and AI agents work together on an equal footing in an iterative process, rather than relying on fully autonomous systems. HAICC addresses key challenges such as automation bias – the uncritical acceptance of AI recommendations – as well as the limited availability of training data in the field of cyber security.

The methods developed are being tested, amongst other things, in scenarios involving the analysis of security vulnerabilities and logs, with AI agents processing heterogeneous datasets and human experts undertaking the evaluation and classification. The project is coordinated by Prof. Dr Iryna Gurevych, holder of the first ATHENE Distinguished Professorship at TU Darmstadt, and will run from 2026 to 2029; participants include nine departments at TU Darmstadt, the Chair of Cyber­security at Goethe University Frankfurt and – as a subcontractor – Ludwig Maximilian University of Munich.

Further information can be found in the press release from TU Darmstadt. (in German)

show all news