Hacking Things That Think - Cognitive Attack Taxonomy and AI Vulnerabilities

Explore the emerging field of cognitive attacks in this 37-minute conference talk from BSidesLV. Delve into the Cognitive Attack Taxonomy (CAT), a comprehensive catalog of over 350 cognitive vulnerabilities, exploits, and TTPs applicable to humans, AI, and non-human biological entities. Examine real-world examples of cognitive attacks, including social engineering techniques, autonomous vehicle disruption, and large language model manipulation. Learn how to create cognitive attack graphs and kill chains for various targets. Witness a demonstration of an ensemble AI model that autonomously assesses vulnerabilities, identifies exploits, and simulates attacks. Gain valuable insights into the similarities between manipulating AI systems and social engineering humans, and understand the implications for cybersecurity in an AI-driven world.