Identity checks before remote exams: matching the person who studied to the person who sits

The main methods and their trade offs

Document plus selfie comparison is the common baseline: the candidate presents a photo ID on camera and a live image, and software or a reviewer compares them and checks the document's basic features. It is fast and familiar, and it can be defeated by determined impersonation or poor image quality, which is why high stakes programmes add liveness checks that confirm a real, present person rather than a held up photo.

Biometric enrolment goes further by binding the account to a reference image captured at registration, then comparing each exam session against it. That catches the classic substitute sitter, where a stronger student takes the exam for the enrolled candidate. It also raises the governance bar: consent, retention limits, data security, and a non biometric alternative route all need to exist and be explained before candidates are asked to enrol.

Design the failure path before you need it

Innocent failures are guaranteed at scale: expired documents, name changes not yet reflected, lighting that defeats a camera, assistive setups that break the standard flow. A programme that treats every verification failure as attempted fraud will punish its most vulnerable candidates and generate appeals it deserves to lose. The failure path, human review, alternative document options, a test centre fallback, should be designed, staffed, and published in advance.

Timing matters too. Verify identity at enrolment or in a practice session days before the exam, not in the ten anxious minutes before it. Early verification surfaces document problems while they can still be fixed calmly, and it shortens exam day check in to a comparison rather than a first time capture under pressure, which is when both errors and tempers peak.

Impersonation patterns to design against

The realistic threats are specific: a paid substitute sitting the whole exam, the enrolled candidate starting the session and swapping out, and document borrowing within families or cohorts. Each pattern has a different control. Enrolment binding helps against substitutes. Periodic in session comparison helps against swaps. Document checks that record and compare across a candidate's history help against casual borrowing that repeats across sittings.

Programmes should be honest about which patterns they actually face, using their own incident data rather than vendor fear. A low stakes internal assessment and a licence to practise examination live in different threat worlds, and controls proportionate to the real one protect candidates and budgets. Overbuilding identity theatre for a low stakes quiz teaches candidates that security is performance, a lesson they carry to the exams where it is not.

Privacy is part of integrity

Identity systems collect the most sensitive data a programme will ever hold, and candidates know it. State plainly what is captured, why, how long it is kept, who can see it, and how to challenge a wrong match. Keep retention short and purpose bound, and delete on schedule rather than on request alone. A breach or a mission creep story in this database does more damage to a programme's legitimacy than any single cheating case it prevented.

Handled well, identity verification is the least resented control in remote assessment, because candidates understand fairness intuitively: everyone proves they are themselves, so results mean the same thing for everyone. Handled badly, with opaque biometrics and no appeal, it becomes the story candidates tell about the programme. The technology is the easy part. The trust design around it decides whether it works.

Keep reading: Lockdown browsers explained: what they can block, what they cannot, and why policy still matters, Brain dump sites and exam content theft: how harvested questions lose their value, and AI tools and remote exams: writing a policy candidates can actually follow cover the nearby cases in this series.

Frequently asked questions