AI Under the Hood: Unmasking Hidden Threats

Learn about AI threat modeling frameworks and their practical applications in this 25-minute conference talk from LASCON. Explore the parallels between safety testing in vehicles and AI technologies, understanding how different AI models require distinct analysis approaches based on risk factors, size, and application domains. Dive into findings from a meta-analysis of 14 AI threat modeling frameworks, examining the challenges posed by inconsistent terminology, varying focus areas, complexity, and interoperability issues. Discover a streamlined approach to AI/ML threat analysis through refined questions developed with expert review, designed to facilitate seamless integration into manual analysis of AI/ML applications.