The value of a credential depends on one thing: the person who earned it is the person who sat the exam. Generative AI has supercharged proxy test-taking. Deepfakes, injected virtual cameras, and synthetic IDs now defeat human proctors and legacy ID checks. This session maps how AI has rewritten the impersonation threat model and offers a practical framework for moving identity from a one-time checkpoint to a continuous, privacy-preserving control: document validation, presence continuity that confirms the same person stays present without storing identity, and human review as a fallback. Scoped to the identity layer, and drawing on the experience of practitioners who sit at the security and test delivery layer. Attendees leave with a threat model, a way to measure identity assurance, and criteria to evaluate solutions, with candidate accessibility built in.