University of Missouri researchers have developed an AI-powered method to detect hidden hardware trojans in computer chips with 97% accuracy. Hardware trojans are malicious modifications embedded in chip designs that can steal data, weaken security, or cause system failures. Unlike software malware, they cannot be removed after manufacturing and often go undetected until activated by an attacker—posing serious risks to devices used in smartphones, medical equipment, finance, defense, and critical infrastructure. Traditionally, identifying these threats has been slow, costly, and complex. The Mizzou team, led by doctoral candidate Ripan Kumar Kundu, has created a new framework called PEARL that uses large language models—similar to those behind popular chatbots—to analyze chip designs. The system not only detects suspicious code with high precision but also provides clear, human-readable explanations for its findings. This transparency helps engineers quickly understand why certain design elements are flagged, eliminating the need to manually review thousands of lines of code. The research, published in IEEE Access, highlights the system’s adaptability. It can operate on local machines or cloud platforms, making it accessible to both open-source developers and large corporations. The method is designed to be integrated into chip design workflows across industries, strengthening security at every stage of the global supply chain. Khurram Khalil, a co-author and fellow doctoral candidate, emphasized the importance of this work. He noted that chips are the backbone of modern digital life, and protecting them from hidden threats is essential. By combining AI with explainable results, the team is creating tools that are not only effective but also trustworthy. Beyond detection, the researchers are exploring real-time correction capabilities to automatically fix vulnerabilities before chips are manufactured. They also believe the approach could be extended to safeguard other critical systems, such as power grids and transportation networks. The development marks a significant step forward in securing the global semiconductor supply chain, offering a faster, more reliable, and scalable solution to a growing threat in the age of advanced AI and interconnected technology.