AI-assisted proctoring has advanced rapidly in remote testing, but physical test centers—where many high-stakes exams are delivered—present different security and operational challenges. As threats evolve, testing organizations are examining how AI can strengthen integrity in person while preserving fairness and candidate trust. Drawing on the design and deployment of AI-assisted monitoring across a global test center network, this session explores the practical considerations involved. Speakers will address model design and false positives, human oversight and centralized monitoring, real-time coordination across distributed locations, and privacy-by-design. Attendees will leave with a practical framework for evaluating AI-assisted security in physical test centers, including measures of impact beyond detection accuracy.