Securing Tomorrow's Hardware - Dual-State Models in the AI Era

Explore hardware security challenges and solutions in the AI era through this 21-minute conference talk that examines dual-state security models for protecting tomorrow's computing infrastructure. Learn about AI's unique vulnerabilities and discover how traditional security approaches must evolve to address the specific threats posed by artificial intelligence systems. Understand address space partitioning techniques that create isolated execution environments for critical AI workloads. Master secure boot processes and runtime integrity mechanisms that ensure AI systems maintain their security posture throughout operation. Discover how AI-driven anomaly detection can be leveraged to identify and respond to security threats in real-time. Examine dual-state security models that provide robust protection by maintaining separate secure and non-secure execution states. Investigate supply chain security considerations specific to AI hardware and software components. Analyze future challenges in hardware security as AI systems become more prevalent and sophisticated, including emerging threat vectors and the need for adaptive security frameworks that can evolve with rapidly advancing AI technologies.