Most discussion of biometric fraud detection focuses on the border — the moment a traveller presents their document and their face to a verification system. But the most effective point of intervention is often earlier, at the moment when the fraudulent document is created. Online identity document issuance is expanding rapidly across Europe and worldwide. It is also a significant attack surface.
The morphing attack on online issuance
A criminal and an accomplice collaborate to create a morphed facial image — a composite blending features of both faces. This image, if accepted as the passport photograph, will be successfully matched by face recognition systems against either contributor. The resulting document becomes a shared identity that both individuals can use to cross borders. The online channel makes this attack easier because the applicant controls the capture environment completely.
Differential detection in the issuance context
Authorities responsible for identity document issuance typically hold previous records — an earlier passport photograph, a national identity register image. This creates the conditions for differential morphing attack detection: comparing the newly submitted image against a trusted reference to identify inconsistencies consistent with morphing. This comparison is more powerful than trying to detect a morph from the submitted image alone.
Quality as a first line of defence
Face image quality assessment serves as an important first line of defence. Morphing algorithms produce images that, while visually plausible, often contain subtle artefacts that manifest as quality issues. A rigorous quality check that goes beyond simple resolution and alignment requirements can flag suspicious images for further review even before a morphing attack detector is invoked.
The EINSTEIN project’s Online Identity Issuance application, developed in partnership with NTNU and the University of Reading with the UK Home Office as end-user, is building and evaluating exactly this layered defence for deployment in real online issuance workflows.
© 2026 EINSTEIN Consortium. EINSTEIN is funded by the European Union’s Horizon Europe programme (GA No. 101121280) and by UKRI (IFS 10093453). Views expressed are those of the authors only. www.einstein-horizon.eu