Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Explore the critical challenges and opportunities in securing AI workloads through trusted hardware in this 20-minute conference talk from USENIX Security '25. Examine how confidential computing and trusted execution environments are being extended from traditional CPUs to clusters of AI accelerators as companies rapidly adopt AI technologies. Learn about the key technical contributions needed from security experts, including efficient remote attestation and key management systems, secure interconnects, and robust device memory protection mechanisms. Discover how these advancements can provide stronger security guarantees while maintaining the performance and code compatibility essential for commercial AI adoption. Gain insights from industry experience in evaluating emerging technologies, offering a comprehensive perspective on both the commercial potential and technical feasibility of implementing trusted hardware solutions for large-scale AI deployments.
