Speaker
Mr
Ruben Missotten
(ERM-KMS)
Description
Automated offensive cybersecurity is rapidly advancing, with AI—especially LLMs—now capable of autonomously discovering and exploiting vulnerabilities.
While AI excels at isolated tasks like fuzzing, phishing, and web penetration testing, it still struggles to execute full multi‑stage attack chains across complex enterprise environments.
Bridging this gap requires realistic benchmarks that capture real adversarial workflows to guide future development and anticipate emerging AI‑driven threats
