Security Challenges in a World of AI Everywhere

Scott Best, Senior Director of Anti-Tamper Technology at Rambus, discusses the security challenges associated with AI chips and their growing deployment at the edge. Best highlights the similarities between AI chips and FPGAs, particularly in their vulnerability to attacks on valuable firmware assets, such as neural net models and RTL. He explores the importance of securing firmware from tampering and theft, emphasizing the need for encryption and authenticity verification. Best also delves into the rising threats posed by side-channel attacks like Differential Power Analysis (DPA) and offers novel solutions such as key hash chaining to protect firmware while maintaining boot efficiency. He concludes by discussing emerging risks, including neural net model extraction, and suggests tailored countermeasures to safeguard AI architectures from these attacks.

Featured Speaker

Scott Best

Technical Director

Scott Best is the director of anti-counterfeiting products and technologies in the Security business at Rambus. His principle areas of expertise are mixed-signal circuit design, advanced memory systems, 2.5D and 3D packaging technology, anti-tamper technologies and random-number generators. He is one of the most prolific inventors in Rambus’ history, a named inventor on over 120 patents worldwide.

Webinar

Security Challenges in a World of AI Everywhere

Watch Webinar

Featured Speaker

Scott Best

Technical Director