The test security market is a collection of different categories rather than one product market: remote proctoring, identity assurance, secure browsers, device detection, fraud forensics, secure print, credential verification, and originality tooling. New expert research adds a procurement warning: a product that monitors the candidate’s main device may not address a separate camera, phone, auxiliary computer, wearable, or discreet feedback channel operating off-device. Buyers should map each product to a defined integrity risk, test it against plausible attacks, and treat supplier claims as hypotheses rather than validation evidence.

This page maps the main supplier and market clusters appearing in Test Community Network material on test security and integrity. It is a buyer-orientation page, not a league table, certification, or endorsement list.

Assessment teams often buy “security” as if it were one capability. In practice, a live proctor, biometric check, secure browser, item-forensics service, device detector, and credential-verification platform answer different questions. Separating those questions makes procurement, validation, and governance more defensible. The latest expert research highlights a further distinction between **device security** and **environmental assurance**. AI-assisted monitoring may analyse the exam laptop effectively while a separate device captures the screen and supplies answers outside the monitored system. The market is also responding to supply-side pressure: Janison, for example, positions its technology suite around secure, reliable, and scalable online exams, while Integrity Advocate frames hybrid AI-human proctoring as a way to safeguard credentials in the era of advanced AI cheating;. Those are market signals, not validation studies.

- **Remote proctoring**: live, recorded, or AI-assisted supervision of a session. - **Identity assurance**: checks that the registered candidate is the person being assessed. - **Environment control**: restrictions on browsers, applications, devices, workstations, and surrounding objects. - **Device detection**: controls aimed at phones, wearables, hidden cameras, earpieces, or other covert tools. - **Fraud forensics**: statistical or behavioural analysis used to identify suspicious patterns. - **Credential security**: protection, issuance, verification, and revocation of physical or digital credentials. - **Threat-model coverage**: the extent to which a product addresses the actual attack routes relevant to an assessment. - **Evidence maturity**: the distinction between vendor positioning, implementation evidence, and independent validation.

The market is crowded and feature overlap is substantial. Many suppliers combine identity checks, browser controls, monitoring, analytics, review workflows, and candidate support, but the combination does not make the underlying controls interchangeable. There is practical consensus that buyers should examine the operating model around a product. Human review, appeal routes, accessibility, data handling, incident recovery, and audit trails matter as much as the presence of an AI or biometric feature;. The new research supports a stronger procurement principle: suppliers should be asked how their products perform against off-device threats, not only how accurately they classify activity on the assessment device.

The main contest is not which supplier has the longest feature list. It is whether a product improves the validity and defensibility of the result in the real assessment context. Vendor-authored claims about accuracy, compliance, privacy, scalability, or reduced friction need to be tested against live implementation data. A second contest concerns detection versus redesign. Some suppliers add more monitoring and analytics, while design-led approaches ask whether the assessment should be changed so misuse is less valuable or more visible. Neither approach is universally correct; the choice depends on the construct, stakes, candidate population, and operating capacity. The new evidence also exposes a difficult boundary: environmental assurance may require intrusive checks if the assessment insists on unaided answers. Buyers should ask whether a redesigned or tool-integrated assessment would achieve a more defensible result than progressively more invasive surveillance.

- buying overlapping tools without a clear threat model - treating supplier claims as independent validation - confusing compliance documentation with improved integrity outcomes - over-interpreting automated flags as proof of misconduct - failing to test off-device and auxiliary-device attack routes - underestimating accessibility, privacy, support, and appeal costs - relying on a single vendor or platform without recovery arrangements - overlooking centre staff, subcontractor, item-bank, or credential-workflow risk - allowing procurement features to drive assessment design

| Market cluster | Typical purpose | Main buyer question | |---|---|---| | **Remote proctoring** | Observe sessions and support incident review | What are the false-positive, review, privacy, support, and off-device blind spots? | | **AI-assisted monitoring** | Triage sessions or detect unusual events | Does automation improve decisions or simply increase the alert queue? | | **Identity and biometric assurance** | Reduce impersonation and proxy testing | How are enrolment failure, accessibility, privacy, fallback, and appeals handled? | | **Secure browsers and controlled environments** | Restrict on-device access | Does the restriction fit the construct, and what off-device routes remain? | | **Device and wearable detection** | Address phones, smart glasses, hidden audio, and covert capture | Is screening proportionate and operationally feasible for the stakes? | | **Fraud forensics and analytics** | Find item exposure, response, centre, or cross-session patterns | What evidence turns a signal into a defensible review decision? | | **Secure print and fulfilment** | Protect papers, certificates, and controlled documents | Where are the real chain-of-custody and issuance vulnerabilities? | | **Credential verification** | Confirm that a credential is genuine, current, and linked to a result | Can the organisation revoke or correct the credential when a result changes? | | **Originality and authorship tooling** | Support review of copied, AI-assisted, or non-original work | Is the output a review signal rather than a verdict? | | **Advisory and assurance services** | Review controls, governance, and evidence | What independent validation or audit trail do they actually provide? |

A university sees a proctoring demo that flags suspicious eye movement. Before buying, it asks whether the tool can see a second phone, a hidden earpiece, or a separate screen; whether candidates with accessibility needs will be disadvantaged; and whether the alert workflow can be reviewed and appealed. Those questions often matter more than the demo itself.

- Overview of market clusters and buyer questions. - Threat-modelling presentation on off-device cheating routes. - Supplier overview of digital exam security features. - Supplier discussion of hybrid AI-human proctoring and credential trust. - Buyer-oriented page on provider evaluation and evidence checking. - Comparison of direct monitoring, browser control, originality validation, and redesign.

The market is crowded, with overlapping claims around ID verification, browser lockdown, face or object detection, analytics, multilingual support, privacy positioning, and review automation;;;;;;. Some suppliers are also moving into integrated trust stories. Janison’s online-exam integrity material presents a bundled technology view, while Integrity Advocate highlights hybrid AI-human proctoring as a credential-protection model. Those claims are useful market signals, but they should be checked against independent implementation evidence and clear statements of what the system cannot see;.

### What is the biggest mistake when buying test security tools? Buying for features before defining the actual threat and the evidence the assessment must produce. ### Are privacy-led suppliers automatically weaker? No. Some lower-intrusion approaches are more proportionate and defensible; the key is whether the control fits the risk. ### Should plagiarism tools sit in the same shortlist as proctoring tools? Usually not, unless the assessment problem is specifically about writing integrity as well as session security. They solve different problems. ### How should I evaluate an AI-assisted proctoring claim? Ask what the AI actually changes, what it cannot see, how flags are reviewed, and what the false-positive burden looks like in live use. ### Do new smart-device threats mean I need more surveillance? Not necessarily. It may mean you need better threat modelling, stronger device controls, or a different assessment design.

Tim Burnett (Admin)

`Test Community Network. "Test security vendor landscape." TCN AI & Assessment Wiki. Last reviewed 2026-10-22. https://testcommunity.network/knowledge/test-security/wiki/test-security-vendor-landscape`

- Overview of market clusters and buyer questions. - Threat-modelling presentation on off-device cheating routes. - Supplier overview of digital exam security features. - Supplier discussion of hybrid AI-human proctoring and credential trust. - Buyer-oriented page on provider evaluation and evidence checking. - Comparison of direct monitoring, browser control, originality validation, and redesign.