Online proctoring glossary

AI proctoring

AI proctoring uses computer vision and audio analysis to flag suspicious moments in an exam so that a human reviewer can decide.

Definition

AI proctoring, also called automated proctoring, uses machine learning to analyse the candidate's video, audio and screen activity during an exam. Typical analyses include face verification, counting the people in view, detecting objects such as phones or books, estimating gaze direction, detecting speech and checking lighting and image quality.

The software does not prove cheating. It produces flags with timestamps and short pieces of evidence, which a person then reviews. Well-designed systems look at duration and repetition rather than a single frame, so that ordinary behaviour such as drinking water or glancing away briefly does not generate a flag.

AI proctoring scales well because one reviewer can handle many sessions, but it raises questions of accuracy, bias and data protection, so institutions should test it, explain it to candidates and keep a human decision step. In QProctor, the AI proctor can also warn the candidate on screen when it sees a suspicious pattern, and the final decision always stays with the institution.

See it in practice

Try QProctor free for 14 days with up to 20 candidates.

Start free