EAB previews biometric injection attack detection standardization developments

The European Association for Biometrics ( EAB ) is shaping up formal standards for ways to detect and classify biometric injection attacks, as the industry gets serious on its approaches to emerging threats to biometric systems. Experts within the Association say work is under way to create a standardized methodology that can level‐set biometric capture environments and provide a common framework for assessing vulnerabilities. Under discussion is a minimum, internationally recognized list of injection attack methods (IAMs). This would give vendors and regulators the same technical language when evaluating risk. Things are moving to an ISO standard , in keeping with the methodology and approach of the EU’s CEN/TS 18099 standard , which is the only currently-operational formal technical specification that proves injection attack resilience. The ISO/IEC 25456 is an international standard for IAD currently in development based on the European standard. The EAB says the new standard is likely to have a weighting approach in measuring the complexity of attack, with different levels. For example, low and simple to increasingly higher-scored for more difficult-to-detect and hard-to-block attacks. The effort builds on early work by Germany’s federal cybersecurity agency, BSI, with the agency and the EAB examining injection attacks in 2021 — well before the biometric industry fully understood their implications. Injection attacks, whether in the form of replay attacks, device emulators or API manipulation are on pace to total more than 4 billion attacks by 2028, as forecast in the 2026 Injection Attack Detection Market Report and Buyers Guide from Biometric Update and Goode Intelligence . The standards push comes as the industry draws more nuanced distinctions between different types of biometric threats. The relation between “presentation attacks,” “spoofing” and injection attacks is one example of increasing market maturity, as discussed by representatives from IAD providers and testing practitioners in a recent webinar from Biometric Update . The distinctions are important. Presentation attacks are generally manual and low‐scale. Injection attacks, EAB experts warn, are far more scalable because once an attacker can push manipulated data into a system, they can repeat the process at volume. Border control systems and other tightly secured capture environments are generally more resistant to presentation attacks, but their increasing reliance on digital transmission channels could make them more exposed to injection‐based threats. Varying risk profiles are driving demand for clearer standards and shared terminology. The EAB says standardization will help the industry better protect biometric systems by ensuring all attack surfaces are addressed and by giving regulators and vendors a common baseline for testing and certification. Explainability and a strategic framework for harmonization EAB experts expect explainability to become a requirement for detecting deepfakes and injection attacks in biometric systems. It’s related to the EU’s regulations on explainable AI. Vendors will likely need to disclose the security methods they use, particularly as standards bodies tighten testing obligations. A “ Practical AI Act Guide ” from the Applied AI Institute for Europe can help vendors wrap their heads around explainability, which can aid compliance with regulation as well as provide insights into a system’s decision-making process. This is perhaps a more general guide but the explainability techniques should inform what could be required regarding biometric systems as compliance measures are still being mapped. Under ETSI EN 119 461, injection attack detection already requires mandatory testing and documented evidence. The standard goes beyond traditional presentation attack detection by explicitly covering injection attack scenarios. Scenarios under ETSI 119 461 v2.1.1 include face-to-face identification, real-time remote assisted identification and unattended remote identification. Improved user experience, harmonized practices and fraud prevention are key aspects of the ETSI. It is a strategic framework as much as it is about technical requirements. On device‐level security, experts say cryptographic attestation at the camera‐firmware level is technically possible but unlikely to become mandatory soon. Many consumer cameras — whether webcams or embedded sensors — remain accessible and unprotected, and market demand rather than regulation is expected to drive adoption. Software‐based behavioral analysis may help detect man‐in‐the‐middle injections, but hardware protections are not yet seen as a near‐term requirement. Generally, the approach of “security by design” is seen as essential. Any modern biometric system should account for injection attack risks from the outset.

The European Association for Biometrics (EAB) is shaping up formal standards for ways to detect and classify biometric injection attacks, as the industry gets serious on its approaches to emerging threats to biometric systems. Experts within the Association say work is under way to create a standardized methodology that can level‐set biometric capture environments and provide a common framework for assessing vulnerabilities. Under discussion is a minimum, internationally recognized list of injection attack methods (IAMs).

This would give vendors and regulators the same technical language when evaluating risk. Things are moving to an ISO standard, in keeping with the methodology and approach of the EU’s CEN/TS 18099 standard, which is the only currently-operational formal technical specification that proves injection attack resilience. The ISO/IEC 25456 is an international standard for IAD currently in development based on the European standard.

The EAB says the new standard is likely to have a weighting approach in measuring the complexity of attack, with different levels. For example, low and simple to increasingly higher-scored for more difficult-to-detect and hard-to-block attacks. The effort builds on early work by Germany’s federal cybersecurity agency, BSI, with the agency and the EAB examining injection attacks in 2021 — well before the biometric industry fully understood their implications.

Injection attacks, whether in the form of replay attacks, device emulators or API manipulation are on pace to total more than 4 billion attacks by 2028, as forecast in the 2026 Injection Attack Detection Market Report and Buyers Guide from Biometric Update and Goode Intelligence. The standards push comes as the industry draws more nuanced distinctions between different types of biometric threats. The relation between “presentation attacks,” “spoofing” and injection attacks is one example of increasing market maturity, as discussed by representatives from IAD providers and testing practitioners in a recent webinar from Biometric Update.

The distinctions are important. Presentation attacks are generally manual and low‐scale. Injection attacks, EAB experts warn, are far more scalable because once an attacker can push manipulated data into a system, they can repeat the process at volume.

Border control systems and other tightly secured capture environments are generally more resistant to presentation attacks, but their increasing reliance on digital transmission channels could make them more exposed to injection‐based threats. Varying risk profiles are driving demand for clearer standards and shared terminology. The EAB says standardization will help the industry better protect biometric systems by ensuring all attack surfaces are addressed and by giving regulators and vendors a common baseline for testing and certification.

Explainability and a strategic framework for harmonizationEAB experts expect explainability to become a requirement for detecting deepfakes and injection attacks in biometric systems. It’s related to the EU’s regulations on explainable AI. Vendors will likely need to disclose the security methods they use, particularly as standards bodies tighten testing obligations.

A “Practical AI Act Guide” from the Applied AI Institute for Europe can help vendors wrap their heads around explainability, which can aid compliance with regulation as well as provide insights into a system’s decision-making process. This is perhaps a more general guide but the explainability techniques should inform what could be required regarding biometric systems as compliance measures are still being mapped. Under ETSI EN 119 461, injection attack detection already requires mandatory testing and documented evidence.

The standard goes beyond traditional presentation attack detection by explicitly covering injection attack scenarios. Scenarios under ETSI 119 461 v2.1.1 include face-to-face identification, real-time remote assisted identification and unattended remote identification. Improved user experience, harmonized practices and fraud prevention are key aspects of the ETSI.

It is a strategic framework as much as it is about technical requirements. On device‐level security, experts say cryptographic attestation at the camera‐firmware level is technically possible but unlikely to become mandatory soon. Many consumer cameras — whether webcams or embedded sensors — remain accessible and unprotected, and market demand rather than regulation is expected to drive adoption.Software‐based

behavioral analysis may help detect man‐in‐the‐middle injections, but hardware protections are not yet seen as a near‐term requirement. Generally, the approach of “security by design” is seen as essential. Any modern biometric system should account for injection attack risks from the outset.