AI Trustworthiness in the era of Advanced Packaging: Challenges and Opportunities

The rapid advancements in artificial intelligence (AI) have driven the adoption of advanced packaging technologies to enhance cost efficiency and energy performance. However, the increasing use of AI in critical sectors like healthcare and finance raises concerns about security, authenticity, and integrity. This work examines the potential threats to AI hardware within advanced packaging technologies, analyzing real-world scenarios where these vulnerabilities could impact critical applications. Furthermore, it explores techniques for establishing trust, including physical assurance-based fingerprinting and watermarking, which offer robust mechanisms for verifying AI hardware integrity. To systematically assess AI trustworthiness, we propose a comprehensive framework that integrates software, physical, and metrology-based metrics. This framework enables continuous monitoring of AI systems for unexpected behaviors, facilitating the detection and mitigation of critical AI trust-related failures.