Vasilios Mavroudis is a Principal Research Scientist and co-lead of the AI for Cyberdefence (AICD) Research Centre at the Alan Turing Institute. His work focuses on systems security, autonomous cyber agents, and AI safety, especially the intersection of machine learning with network defence and offensive cyber capabilities.
His recent efforts include the AI Cyber Risk Benchmark for automated vulnerability exploitation, and authorship of the cyber offense section in the International Report on AI Safety (with Yoshua Bengio et al.). His research appears in venues like NeurIPS, CCS, PETs, and NDSS, and he's a frequent speaker at both academic and hacker conferences (BlackHat, DEF CON, CCC).
In past lives: organised hacking and lockpicking seminars at UCL, built mobile security systems in Greece, and earned a nod from the Heidelberg Laureate Forum.
Contact: vmavroudis@turing.ac.uk
More: Website | Google Scholar | Group
Interesting article! Here is a benchmark that does real world exploitation (indeed CTFs don't cut it anymore): https://github.com/alan-turing-institute/HonestCyberEval