Repository navigation
Topic G - Zero Knowledge Proofs (Refinement Round) #700
Replies: 1 comment
|
Three requirements from financial crime prevention: third-party lists, the proof as the record, and request-side unlinkability Thank you for reopening Topic G. I research zero-knowledge proofs for anti-money-laundering checks by obliged entities in payments and crypto-asset services, including stablecoin issuers. From 10 July 2027 the Anti-Money Laundering Regulation, Regulation (EU) 2024/1624, lets obliged entities verify identity through electronic identification means at assurance level substantial or high under Article 22(6)(b). The EUDI Wallet then becomes a lawful credential source for customer due diligence. From that use case I suggest three additions to ZKP_01 to ZKP_09 in ARF 3.0.0. Each comes with the situation that creates the need. 1. Non-membership in a list kept by a third partyThe situation. A company paying a refund, a prize or a payout has no duty to identify the payee. It is still bound by the prohibition on making funds available to a listed person, for example Article 2(2) of Council Regulation (EU) No 269/2014, which binds everyone whatever the amount. Today it screens the payee's name, which means collecting a name it has no other reason to hold. The fact it needs is one bit, not listed as of today. Why it matters beyond payouts. Section 4.1 of the Topic G paper notes that where full identity verification is required, such as opening a bank account, unlinkability is not possible. The reason is screening. If attribute-based verification is ever permitted for customer due diligence, screening is the step that still demands the name. This requirement therefore decides whether the financial sector can use ZKP at all. The proposal. ZKP_01(iii) already demands a non-revocation proof. That is the same primitive, a non-membership proof against an accumulator published by a party other than the Attestation Provider. I suggest generalising it.
Alternative and limit. A screening attestation issued by a screening provider fits the current requirements, but it creates a new party that learns who is screened and when, and it must be reissued at each list change. The non-membership proof has a limit of its own. It matches exact identifiers such as passport or national identification numbers, not the fuzzy name matching that screening practice uses. It complements a screening attestation rather than replacing it. 2. The proof as the recordThe situation. Article 77(1)(a) AMLR requires obliged entities to retain a copy of the documents and information obtained in customer due diligence, including information obtained through electronic identification means. Article 77(3) sets the retention period at five years. If a relying party receives only a proof, the proof transcript is the record. The proposal. Two requirements follow.
Because retained proofs can be re-examined years later, hiding must also hold against future verifiers. This supports the everlasting-privacy requests made in the 2025 thread. Open legal question. Whether a proof transcript counts as "information obtained" under Article 77 is a question for the AMLA technical standards. The ARF could usefully flag it. 3. Presentation requests about named third partiesThe situation. Consider a platform whose Users register with a ZKP_04 pseudonym and no identity. An authority investigating account X at that platform learns only the pseudonym. If presentation requests may carry any predicate, the authority can have the platform, or any relying party that cooperates with it, ask each User at the next login to prove that their pseudonym for that context differs from X. Every User but one can comply, and refusal marks that User. The pseudonymity that ZKP_04 was designed to give is undone by one request to the whole population, with no order against any named person. Private relying parties have a plausible version of the same request, proving that a new User is not the holder of a banned account. That turns the Wallet into a cross-account identity oracle. Why the current requirements do not cover it. ZKP_07 concerns what a proof reveals. RPRC_21 makes the Wallet Unit check requested attributes against the registration certificate, but it checks attributes, not predicates. "Over 18" and "not the holder of X" can be predicates over the same attribute. The proposal.
A rule at ARF level protects everyone at once, because a request that every Wallet Unit refuses identifies no one. Relation to point 1. A non-membership proof is safe only against a list established by law, identical for all Users and published by a designated authority, never against a set chosen by the requester. A list of one is a targeted unmasking. I am glad to expand on any of these points. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Welcome to the discussion on the Digital Signature using the EUDI Wallet, part of the ongoing development of the Architecture and Reference Framework (ARF) for the European Digital Identity (EUDI) Wallet. This topic has end of December 2025 been assessed to be further refined in the ARF Refinement Round in 2026.
Objective of Topic G - Zero Knowledge Proofs (Refinement Round):
Explore and clarify foundational requirements for future integration of ZKP-based mechanisms in alignment with ongoing technical work.
Description:
This topic explores the potential use of ZKP as privacy-enhancing technique within the EUDI Wallet, including possible implementation approaches using different types of secure cryptographic devices or architectures.
Reason why Topic G is addressed in refinement round:
While ZKP offers significant long-term potential, there is currently no clear, certifiable approach agreed at EU level. Further conceptual work and alignment are needed, including consultation with cryptography experts and relevant bodies such as ECCG. Refinement at this stage focuses on establishing common understanding and boundaries, rather than immediate deployment to secure and privacy-preserving future integration.
We refer to the ARF 3.0.0, including annex 2 on the latest status.
This discussion is part of a structured process to refine and complete the ARF, with your input playing a vital role. We invite you to share your comments, insights, and suggestions here. Your contributions will be carefully reviewed and considered as we work towards the next version of the ARF, which will incorporate updates on this topic.
Thank you for participating in this important conversation.
All reactions