FaceTec’s Level 1 to 5 liveness and injection testing puts remote identity verification’s hardest problem into sharper focus
FaceTec has completed a full suite of third-party Level 1 to 5 liveness and injection-attack testing for its 3D Liveness technology, marking a significant milestone in how biometric liveness is evaluated for remote identity verification. According to the company, the testing resulted in zero successful attacks across all evaluated levels, including advanced injection scenarios designed to bypass the camera pipeline.
While liveness has long been a foundational control in digital onboarding and authentication, the threat environment it must address has changed. Early liveness systems were primarily designed to stop basic presentation attacks such as printed photos or replayed videos. Today’s attackers increasingly target the software layer itself, attempting to replace or manipulate camera input before it ever reaches a liveness engine. FaceTec’s emphasis on Level 4 and Level 5 testing reflects that shift.
“The results from these independent evaluations reaffirm that liveness security today must extend far beyond presentation-attack detection,” said Manos Mavrikos, Senior Vice President of Growth at FaceTec. “Threat actors are no longer just spoofing cameras—they’re targeting the software stack, the device environment, and the biometric pipeline itself.
“By successfully completing high-assurance tampering and injection-attack testing across multiple labs and platforms, FaceTec has again demonstrated that its 3D Liveness architecture is purpose-built for real-world adversaries, not theoretical ones. This is the level of assurance regulators, enterprises, and critical-infrastructure programs now require, and it’s exactly what we deliver.”
In FaceTec’s Level framework, Levels 1 through 3 focus on traditional presentation attacks, validating that a system can reliably distinguish a live human face from physical artifacts. Levels 4 and 5 extend that assurance into software integrity, addressing whether attackers can tamper with capture data, inject synthetic imagery, or otherwise compromise the biometric workflow. The goal is not only to detect fake faces, but to ensure the system is receiving authentic, untampered camera input.
Why injection attacks now matter as much as liveness
Remote onboarding has expanded rapidly in regulated environments, including financial services, government services, and digital identity programs. At the same time, advances in deepfake generation, emulator tooling, and mobile device manipulation have lowered the barrier for sophisticated attacks. Rather than presenting a fake artifact to a camera, attackers can attempt to insert synthetic images or video directly into the capture process, bypassing many of the cues traditional liveness checks rely on.
This evolution has driven growing attention to injection attacks within standards and testing communities. Guidance discussed at the NIST Identity Fraud Prevention Conference has highlighted injection as a primary threat to modern remote identity verification, distinguishing it from classic presentation attacks and emphasizing the need to evaluate the full acquisition and processing workflow. In that context, liveness is no longer just a classification problem. It is also a system-integrity problem.
FaceTec’s Level 4 and Level 5 testing is designed to address this broader threat model by combining biometric analysis with protections against tampering, reverse engineering, and injection at multiple points in the capture and processing flow.
What the Level 4 and Level 5 evaluations covered
FaceTec’s testing involved multiple independent security and biometric laboratories, each focused on different aspects of the attack surface. The goal was to validate that FaceTec’s 3D Liveness can withstand both traditional presentation attacks and attacks that target the software environment itself.
“The Level 4 and Level 5 evaluations are a powerful confirmation of how rigorously FaceTec’s technology is scrutinized—far beyond traditional presentation-attack testing,” said Mavrikos.
“These assessments validated our resistance to software tampering, reverse engineering, payload manipulation, emulator-based execution, virtual-camera injection, API interference, deepfake and morph injection attempts, and rooted-device manipulation.
“Multiple independent labs pushed every conceivable attack vector across Android, iOS, and web, yet none were able to compromise the integrity of our 3D Liveness or the authenticity of the capture pipeline. This depth of testing underscores the maturity of our security architecture and demonstrates why FaceTec continues to set the global benchmark for real-world biometric resilience.”
Level 4 testing was conducted by Praetorian Security and focused on bypass and tampering resistance. The evaluation examined whether an attacker could force a successful liveness outcome or manipulate the integrity of 3D FaceScan data through reverse engineering, payload modification, or other forms of software tampering. The assessment involved extensive hands-on testing and reported no critical, high, medium, or low findings, supporting FaceTec’s claim that its capture and encryption mechanisms are resistant to direct manipulation.
Level 5 testing addressed injection attacks and was carried out by BixeLab and Ingenium Biometric Laboratories. These evaluations simulated scenarios in which synthetic or manipulated imagery is injected into the biometric workflow rather than presented to a physical camera.
BixeLab’s testing covered Android, iOS, and web implementations of FaceTec’s SDK and included a range of attack methods such as virtual camera injection, emulator-based execution, rooted-device manipulation, API injection, and JavaScript tampering. The injected content included selfie images, passport-style images, face morphs, and deepfake videos. Across these scenarios, BixeLab reported that all injection attempts were rejected by the system.
Ingenium’s evaluation similarly focused on injection attacks against an Android implementation, examining emulator-based attacks, function hooking, and rooted-device manipulation. Ingenium reported that none of the tested injection techniques succeeded and characterized the testing as aligned with high-assurance injection attack detection concepts reflected in emerging technical specifications.
“Ingenium’s testing in particular represents exactly the type of rigorous, standards-aligned scrutiny the industry should expect for high-assurance injection-attack detection,” said Mavrikos.
“Their team conducted a full IAD assessment of the FaceTec 3D FaceScan application, following the methodology and requirements of CEN TS 18099, with the Android mobile app as the target of operation.
“The evaluation was performed under Ingenium’s Level 2 criteria—equivalent to a CEN TS 18099 HIGH evaluation—and applied four distinct injection attack methods based on emulators, function hooking, and rooted phones, combined with fifteen different attack-instrument species spanning 2D and 3D imagery, videos, and deepfakes.
“Ingenium created numerous attack types by pairing IAMs with IAI species and weighting them by sophistication and complexity, ensuring the assessment mirrored real-world attacker behaviour. Across a full 30 FTE days of structured testing effort, the evaluation demonstrated the depth, completeness, and robustness expected from a modern high-assurance injection-attack test regimen.”
Taken together, the evaluations demonstrate FaceTec’s approach to treating liveness as an end-to-end security problem. Rather than isolating facial analysis from its execution environment, the testing assesses whether the biometric system can maintain integrity even when attackers target the software stack.
What the results demonstrate for deployers
For organizations deploying remote identity verification, the significance of this testing lies in what it says about resilience under realistic adversarial conditions. FaceTec’s results show that its 3D Liveness technology has been evaluated not only against common spoofing techniques, but also against attacks that attempt to subvert the capture process itself.
This distinction matters as regulators and enterprise buyers become more sophisticated in how they evaluate liveness claims. Increasingly, procurement teams are asking how a system behaves when run in emulators, on rooted devices, or in environments where camera input can be intercepted or replaced. Level 4 and Level 5 testing provides a framework for answering those questions in a structured, repeatable way.
The testing also aligns with broader industry efforts to modernize liveness assurance. Programs such as FIDO’s Face Verification Certification are explicitly incorporating threats such as deepfakes and injection attacks into their scope, reflecting a consensus that future-proof liveness must address both biometric accuracy and workflow integrity.
Positioning liveness for the next phase of digital identity
FaceTec’s completion of Level 1 to 5 testing also points to a broader shift the industry is beginning to confront. As attack techniques evolve, isolated or partial testing is no longer sufficient to keep pace. Presentation attacks, software tampering, and injection attacks are not separate problems that can be addressed independently; they represent a continuum of adversarial behavior that adapts as defenses improve.
In that context, full-spectrum testing across all five levels is not simply a signal of readiness for today’s threats. It is a practical mechanism for containing the growth of tomorrow’s attack surface. By forcing vendors to validate claims across presentation, tampering, and injection scenarios in a structured and repeatable way, comprehensive testing helps ensure that new attack techniques are met with corresponding defensive advances, rather than creating blind spots that attackers can exploit.
This dynamic is increasingly relevant as remote identity verification becomes embedded in higher-assurance use cases. Attacks are evolutionary by nature, and defenses that are rigorously tested against known techniques tend to be better positioned to absorb the next wave of innovation on the attacker side. When vendors are required to substantiate their claims through independent evaluation, the industry as a whole benefits from a more stable security baseline.
Conversely, when liveness claims are not held accountable in a sanctioned, transparent way, the cost of defending identity systems rises sharply. Gaps between marketing assertions and real-world resilience can allow entire classes of attacks to scale before they are properly addressed, making remediation more complex and more expensive for relying parties.
From that perspective, FaceTec’s Level 1 to 5 testing is not just about demonstrating the strength of a single product. It reflects an emerging expectation that liveness solutions should be evaluated holistically, against the full range of attacks that modern identity systems face. As digital identity continues to expand, comprehensive, multi-level testing is likely to become a baseline requirement for maintaining trust, controlling risk, and preventing the attack landscape from escalating faster than defenses can respond.
–
By Cass Kennedy and Alex Perala
