Online proctoring is the set of methods and technologies used to supervise a remote exam and confirm that the person completing it is who they claim to be and is not using unauthorized resources. Three main modes exist: live proctoring, where a human proctor watches via webcam in real time; record-and-review, where the session is captured and a human reviews it later; and automated proctoring, where AI flags anomalies without a human in the loop. Each mode represents a different trade-off among cost, scale, privacy intrusion, and accuracy. Live proctoring offers nuanced human judgment but is expensive and difficult to scale; automated proctoring is cheap at scale but produces false positives that can unfairly penalize learners. Proctoring necessarily involves collecting sensitive data, including webcam footage, screen recordings, and sometimes biometric signals, which creates legal obligations under frameworks such as GDPR and FERPA and requires transparent consent. Assessment design is also a powerful deterrent: open-book formats, unique questions, time pressure, and scenario-based tasks can reduce the value of cheating without invasive surveillance. Vendors implementing proctoring systems must consider data minimisation, storage duration limits, and learner appeal processes as first-class requirements, not afterthoughts.
