University of Idaho Team Wins Global AI Chip Security Challenge #AI

DAC brings together more than 5,000 researchers, engineers, industry professionals and students to explore the future of semiconductor technology, artificial intelligence and hardware design. As an undergraduate, Alcorn competed alongside teams from academia and industry.“It was a chance to get my feet wet as an undergraduate student, and I had no idea experiences like this were even possible until I came to U of I,” Alcorn said.The competition ran from June to July 2026. Students register to participate as red team members, acting as attackers and blue team members ready to defend their systems. Teams meet their objectives independently then submit final results for demonstration at the conference.DAC brings together more than 5,000 researchers, engineers, industry professionals and students to explore the future of semiconductor technology, artificial intelligence and hardware design. Alcorn and Abideen created a hardware security system embedded within a microchip to obfuscate, or hide, critical information about the chip’s functionality and intellectual property. They also had to build the system without sacrificing the chip’s computing power or performance. The competition was held remotely, with awards presented during the conference.Microchips made of semiconductor material pass through multiple manufacturing, packaging, testing and distribution facilities, creating opportunities for sensitive information about a chip’s design and capabilities to be exposed or stolen.“Students entering the semiconductor field need to understand both how advanced chips are designed and how the intellectual property inside them can be protected,” Abideen said. “This competition gave Lloyd hands-on experience with generative AI, chip-design tools and adversarial security analysis on a real-world problem. That combination is rarely available at the undergraduate level.”Over several summer weeks, Alcorn adapted and tested the protection workflow, verifying that the protected designs continued to operate correctly, and measuring the effects on chip power and timing. He also conducted an AI-assisted adversarial evaluation to determine what information an attacker might still be able to learn from the protected design.The remote summer competition pitted both sides of hardware security against each other. Blue teams developed methods to conceal sensitive design information, while red teams used generative AI to test whether that information could be reconstructed. “Research experience at the intersection of hardware design and security requires experienced faculty willing to be patient with someone still learning. That’s what I found at U of I,” Alcorn said.

Click Here For The Original Source.

——————————————————–

…………

Related