Purpose
PAD is a biometric control layer; claims are bounded to certificate scope.
PAD is a biometric control layer; claims are bounded to certificate scope.
PAD is a biometric control layer; claims are bounded to certificate scope.
PAD is a biometric control layer; claims are bounded to certificate scope.
Capability is published only within source/evidence scope.
SDK · REST API · evidence contract
↗ISO/IEC 30107-3:2023 focuses on PAD performance assessment and reporting. It does not prescribe a specific anti-spoofing algorithm or overall system security.
Legitimate presentations used to evaluate normal-user behaviour.
A presentation intended to interfere with or deceive the capture/PAD subsystem.
Evaluates capture evidence within the assessed component boundary.
Results must remain bound to test protocol, attack set, platform, version and certificate scope.
The current reference edition and the certificate basis must be shown separately.
| Item | Meaning | Publication rule |
|---|---|---|
| APCER | Attack presentations classified as bona fide | Publish numbers only with matching report/evidence |
| BPCER | Bona fide presentations classified as attack | Not equivalent to face-match accuracy |
| ISO/IEC 30107-3:2023 | Current PAD testing/reporting reference | REFERENCE |
| ISO/IEC 30107-3:2017 | Basis stated on PCEB 26/04/06 | CERTIFICATE BASIS |
| PAD Level 2 | Level in the applicable assessment/certification scheme | Do not describe as a universal ISO product level |
Presentation attack ≠ capture injection ≠ client compromise ≠ API attack
ISO/IEC 30107-3:2023 addresses PAD performance testing and reporting for attacks occurring at the biometric capture device during presentation. It explicitly does not constitute an overall system vulnerability assessment.
Virtual camera, injected image/video and emulator paths may bypass normal capture and are not automatically equivalent to presentation attacks. CheckID anti-injection controls are claimed only with security architecture or test evidence.
Root/jailbreak, application hooking and payload tampering belong to the client-integrity layer. A PAD certificate does not establish these controls.
Replay, credential theft, request substitution and API misuse belong to channel/API security. Authentication, request signing, token handling and replay controls require separate evidence.
Stolen evidence, document substitution, synthetic identity and source conflict are broader fraud scenarios that belong in proofing/risk/source policy rather than being labeled liveness attacks.
PCEB 26/04/06 remains bound to the ISO/IEC 30107-3:2017 basis, Level 2, module/platform scope and validity stated on the certificate. The current 2023 technical reference does not rewrite the certificate basis.
The threat taxonomy separates presentation attacks at capture from injection, client compromise, API/channel attacks and identity fraud so ISO/IEC 30107-3 or a PAD certificate is not stretched beyond scope.
| Layer | Examples | Evidence boundary |
|---|---|---|
| Presentation | Print, display replay, mask/presentation artefact | ISO/IEC 30107-3/certificate only for the evaluated module/platform/scope. |
| Capture injection | Virtual camera, injected media, emulator path | Requires separate security/test evidence; not inferred from PAD. |
| Client compromise | App tampering, hooking, rooted/jailbroken | Controls are published only from mobile/client security sources. |
| Channel/API | Replay, credential abuse, request substitution | API authentication does not prove all replay controls. |
| Identity/evidence fraud | Stolen evidence, source conflict, impersonation | Requires corresponding proofing/source/policy controls. |