How to Benchmark Liveness Detection Against iBeta Level 1 and Level 2

The ISO/IEC 30107-3 standard for Presentation Attack Detection (PAD) provides a critical framework for evaluating the security of biometric systems. As identity verification vendors and enterprise fraud teams navigate an increasingly complex threat landscape, understanding how to benchmark liveness detection iBeta Level compliance has become essential for risk management and technology procurement. Third-party testing, particularly through accredited laboratories like iBeta, offers a standardized measure of a system's ability to thwart spoofing attempts. This process is not merely a technicality; it is a foundational component of building trust in digital identity frameworks, ensuring that the person on the other side of a transaction is physically present and real.

"A Presentation Attack is an action of presenting a biometric artefact to a biometric capture subsystem with the goal of interfering with the operation of the biometric system. The ISO/IEC 30107 standard series was developed to provide a foundation for Presentation Attack Detection."

• Stephanie Schuckers, Director of the Center for Identification Technology Research (CITeR), 2018

Benchmarking Liveness Detection: iBeta Level 1 vs. Level 2

iBeta is an independently accredited testing laboratory that specializes in quality assurance, including Presentation Attack Detection testing for biometric systems. The lab tests liveness detection solutions against the ISO/IEC 30107-3 standard, providing a clear benchmark for their effectiveness. Achieving compliance involves a rigorous process where a biometric solution is subjected to a barrage of spoof attempts using various artifacts. The goal is to determine the system's Attack Presentation Classification Error Rate (APCER), the rate at which it incorrectly accepts a spoof as a live person, and its Biometric Presentation Classification Error Rate (BPCER), the rate at which it incorrectly rejects a live person.

To achieve compliance, a system must demonstrate an APCER of 0% across thousands of test attempts. The distinction between Level 1 and Level 2 lies in the nature and complexity of the presentation attacks used during testing. This distinction is critical for organizations to understand when selecting a liveness detection vendor, as the level of compliance directly correlates to the types of threats the system is proven to mitigate. For any organization needing to benchmark liveness detection iBeta Level compliance, understanding these tiers is the first step.

Feature	iBeta PAD Level 1	iBeta PAD Level 2
Governing Standard	ISO/IEC 30107-3	ISO/IEC 30107-3
Primary Goal	Detect presentation attacks using basic, easily accessible artifacts.	Detect presentation attacks using sophisticated, well-crafted artifacts.
Attack Artifacts	High-quality printed photos, high-resolution digital screen replays.	Custom-made 3D masks (silicone, latex, resin), hyper-realistic digital avatars, deepfake videos.
APCER Requirement	0%	0%
BPCER Requirement	Defined by vendor (e.g., < 15%)	Defined by vendor (e.g., < 15%)
Effort Level	Low-effort attacks that an average individual could attempt.	High-effort, skilled attacks requiring significant resources and expertise.
Test Report Validity	Typically 12-24 months, requires re-testing as technology evolves.	Typically 12-24 months, requires re-testing.
Common Use Case	Low-to-medium security applications.	High-security applications (banking, government ID, enterprise security).

Industry applications and compliance requirements

The need for robust PAD testing is not uniform across all industries. The level of assurance required often depends on the financial, social, or security-related risks of a fraudulent transaction. As regulators and industry bodies create more formal requirements, iBeta compliance is shifting from a best practice to a contractual necessity.

### financial services and fintech

For banks, neobanks, and KYC providers, preventing account opening fraud is a primary driver for adopting liveness detection. The risk associated with a compromised account is high, making robust spoof detection critical. Many financial regulations worldwide now implicitly or explicitly require strong customer authentication that can resist sophisticated spoofing attempts. Therefore, iBeta Level 2 compliance is rapidly becoming the de facto standard for vendors serving this market. It provides documented, third-party assurance that the system can defend against the high-effort attacks expected from organized fraud rings.

### government and public sector

Government agencies providing digital access to benefits, issuing digital IDs, or securing voting systems require the highest level of assurance. The scale of these systems means that even a small vulnerability can have widespread consequences. Level 2 compliance is essential for these use cases, providing confidence that the biometric systems can withstand attacks from determined adversaries, including state-sponsored actors.

### enterprise security and access control

Enterprise use cases, such as securing employee access to sensitive data or physical locations, also benefit from PAD testing. While Level 1 might suffice for lower-risk internal applications, systems protecting critical infrastructure or high-value intellectual property should be benchmarked against Level 2 standards. The ability to thwart attacks from realistic 3D masks is particularly relevant for physical access control.

Current research and evidence

The field of Presentation Attack Detection is in constant evolution, driven by an arms race between security researchers and malicious actors. Current research, much of it presented at conferences like the IEEE International Conference on Biometrics: Theory, Applications and Systems (BTAS), focuses on several key areas.

• Novel Attack Vectors: Researchers like Stephanie Schuckers at Clarkson University and Sébastien Marcel at the Idiap Research Institute have published extensive work on new types of presentation attacks. Their research from 2020 onward explores the vulnerabilities of biometric systems to deepfakes and highly realistic 3D masks, pushing the boundaries of what standards like ISO/IEC 30107-3 need to cover.
• Physiological Signals: A growing body of research supports the use of physiological signals, such as those captured by remote photoplethysmography (rPPG), as a more robust basis for liveness detection. A 2022 study by Unal et al. demonstrated that algorithms analyzing micro-level blood flow patterns in the face could successfully differentiate between live subjects and sophisticated digital or physical spoofs. This approach is inherently resistant to attacks that merely mimic appearance, as they cannot replicate the physiological signature of a living person.
• Standardization Gaps: While ISO 30107-3 is the gold standard, researchers note that it is a point-in-time evaluation. The standard itself does not account for zero-day attacks or the rapid evolution of generative AI. This has led to calls for more adaptive testing frameworks and a move toward continuous monitoring and re-evaluation of certified systems.

The future of liveness detection benchmarking

The future of PAD testing will be defined by the need to keep pace with AI-driven threats. As generative AI makes it easier and cheaper to create high-fidelity spoofs, the distinction between Level 1 and Level 2 may become less meaningful. The industry is moving toward a paradigm where all publicly accessible systems must be resilient against sophisticated, AI-generated attacks.

Future benchmarks will likely incorporate new metrics beyond APCER and BPCER, focusing on a system's ability to detect novel or "zero-day" attack vectors. We may see the emergence of "Level 3" testing, which could involve red-teaming exercises where testers use recent generative AI models to probe for weaknesses in real-time. Furthermore, the focus will shift from single-frame analysis to a holistic, time-series evaluation of a subject's interaction, analyzing not just appearance but subtle physiological and behavioral cues that are difficult to synthesize.

Ultimately, the goal is to create a testing environment that accurately reflects the capabilities of modern adversaries. This requires a collaborative effort between standards bodies like ISO, testing labs like iBeta, and the research community to ensure that the benchmarks for liveness detection remain relevant and effective.

Frequently asked questions

What is the difference between active and passive liveness detection? Active liveness detection requires the user to perform a challenge, such as blinking, smiling, or turning their head. Passive liveness detection analyzes the user in the background without requiring any specific action, offering a more seamless user experience. iBeta testing can be applied to both types of systems.

How often should a liveness detection system be re-tested for compliance? iBeta compliance is not permanent. Given the rapid evolution of attack vectors, most experts and procurement teams recommend that vendors undergo re-testing every 12 to 24 months to maintain their compliance status and ensure continued protection against the latest threats.

Can a system be compliant with both Level 1 and Level 2? Yes. A system that achieves Level 2 compliance has by definition been tested against the sophisticated artifacts of Level 2 as well as the simpler artifacts of Level 1. Therefore, Level 2 compliance is inclusive of Level 1.

As the landscape of presentation attacks evolves, organizations are seeking more robust verification methods. Circadify is at the forefront of this challenge, developing next-generation liveness detection based on physiological signals that are inherently resistant to spoofs. To learn how rPPG-based solutions can help your organization meet and exceed compliance standards, explore our solutions for enterprise security at circadify.com/solutions/fraud-detection.