Autonomous Security Analysis & Penetration Testing Frameworks
Dr. Ankur Chowdhary discusses autonomous penetration testing frameworks using attack graphs and reinforcement learning to plan and execute attack paths.
Summary
The Human Bottleneck in Penetration Testing
Manual penetration testing requires extensive human expertise and time, making comprehensive, frequent assessments across expanding enterprise networks difficult.
Developing Autonomous Penetration Testing Engines
Dr. Ankur Chowdhary explores using attack graphs and Markov Decision Processes (MDPs) to model network state transitions and automate exploit planning.
Applying Reinforcement Learning for Adaptive Attack Paths
The speaker demonstrates how autonomous agents learn optimal multi-step attack strategies by probing simulated networks and adapting to defensive responses.
Evaluating the Future of Automated Cyber Defense and Offense
The presentation discusses the integration of autonomous testing tools into continuous purple teaming and security validation pipelines.
Related talks
Cloud Village·25 min
Shani Peled
Exploiting AWS SageMaker to gain persistence and exfiltrate data from ML platforms.
Village Interviews·12 min
An interview exploring WhiteRabbitNeo, an open-source offensive security and red-teaming large language model demonstrated at DEF CON 32.
AI Village·58 min
Vahid Behdazan
Explores vulnerabilities and adversarial manipulation techniques in deep reinforcement learning agents and environments.