The Patents Under Review
PatentReview focuses on the technical and legal analysis of two closely related European patents that are in force in Hungary:
E038266 – the Hungarian validation of European Patent EP 3183680 B1.
E058164 – the Hungarian validation of European Patent EP 3355224 B1.
Both patents belong to the same patent family and seek to address the same technical problem: how to improve the security of electronic document signing and digital authentication by incorporating biometric verification.
The problem the patents seek to solve
An increasing number of everyday activities are now carried out electronically, including signing contracts, authorising banking transactions and verifying our identity online.
Conventional electronic signature systems rely primarily on cryptographic keys and digital certificates. The patents under review extend this process by incorporating biometric authentication, enabling the system to verify the user's identity using biometric characteristics such as a fingerprint or facial image.
How the patented system works
In simplified form, the patents describe the following process:
The user initiates an electronic signature or authentication request.
The server generates a challenge that includes, among other things, a projection parameter.
The client device transforms the user's biometric data using that projection parameter.
The transformed biometric data is transmitted back to the server.
The server verifies the transformed biometric data and, based on the result, approves or rejects the electronic signature or authentication request.
Why are these patents important?
The two patents relate to biometric electronic signatures and electronic authentication, an area that has become increasingly important in information technology and cybersecurity.
According to publicly available information, the commercial exploitation of these patents has generated significant economic value, and substantial royalty payments have reportedly been made to the inventors. In light of this, it is particularly important that the validity of the patents is determined not by speculation or public debate, but through official proceedings conducted in accordance with applicable law and based on technical and legal evidence.
We believe that, for patents of this significance, it is in the interests of the inventors, the patent proprietors, the professional community and the public alike that a transparent process establishes whether the patents satisfy the statutory requirements for patentability. If they do, the resulting decision reinforces the legitimacy of the patents. If they do not, the legal system provides mechanisms for their revocation.
What are we examining?
PatentReview is not assessing whether biometric electronic signatures are, in general, useful or technically feasible.
Our review is limited to determining whether the specific patents E038266 and E058164 satisfy the legal requirements for patentability.
The PatentReview approach
Our objective is not to predetermine the outcome of the proceedings.
In the following sections, we explain step by step:
what the patents claim;
the technical solutions they describe;
the prior art on which they rely;
the technical and legal issues that arise; and
the evidence supporting our analyses.
The final decision, in every case, can only be made by the competent authority or the courts.
Illustrating the Difference Between Lack of Inventive Step and Insufficient Disclosure
In our revocation petition, alongside the objections under Article 56 EPC, we also rely on insufficient disclosure under Article 83 EPC. The following analogy illustrates the difference between these two legal grounds.
Imagine a fruit pie recipe that lists the necessary ingredients — fruit, flour, sugar, cinnamon — but does not specify which fruit is meant: apple, pear, cherry, or plum. The water content, acidity, and thickening requirements of these fruits differ substantially, so without this information the recipe fails to answer even the most basic question: what kind of dough, what kind of thickener, and what preparation technique should be used. The recipe correctly does not elaborate on the technique of making the dough or processing the fruit — this is a routine step, well known to any experienced home baker (common general knowledge, CGK), and omitting it is not, in itself, a deficiency.
But even if the baker were to somehow pick a fruit, the recipe still would not state the correct ratio of flour to fruit. This ratio is not something that could be reliably determined through routine trial and error with a limited number of attempts: the correct ratio also depends on further, likewise unstated variables — how many pies are being baked at once, and how sweet or sour the particular fruit is — so determining it would require independent recipe development, i.e. the baker's own professional effort going beyond what the recipe teaches.
The patent at issue suffers from exactly the same twofold defect, and to an even greater degree.
The description treats "biometric data" — the counterpart of the recipe's "fruit" — uniformly, without regard to modality, even though the description itself lists, as mutually distinct embodiments, handwritten signature dynamics, iris images, fingerprints, facial images, and even mouse-movement-based data, without claim 1, or any other claim, specifying which modality the method actually concerns. The mathematical structure of these modalities — their dimensionality, distribution, and redundancy — differs fundamentally from one another, so a single, modality-independent projection rule cannot conceivably work equally well across all of these embodiments. The description, then, just like the recipe, does not even disclose "which fruit" is meant.
In addition — again just like the recipe — the description correctly does not elaborate on routine, well-known steps (such as hash generation or XOR-based data association), since these form part of the skilled person's common general knowledge. However, the one element that is essential to the invention and cannot be supplied by CGK — the correct choice of the projection's output dimension, and its lower and upper bounds — is nowhere disclosed, and this gap is made even more severe by the fact that the modality itself is left undefined: the correct dimension depends, modality by modality, on further unstated variables (the number of samples to be distinguished, the permissible degree of distortion, and the native dimensionality and structure of the input data), so determining it would require independent mathematical insight rather than routine experimentation.
Consequently, the patent's description — just like a recipe that names no fruit and gives no ratios — does not enable the skilled person to reliably and systematically carry out the invention across its full claimed scope without exercising their own inventive activity. A working embodiment may occasionally be arrived at by chance, but that would be the result of the skilled person's own effort beyond the teaching of the description, not of the description itself — and this is precisely the legal standard for insufficient disclosure under Article 83 EPC