European Digital Identity Wallet – standards and technical specifications (update)

European Commission · Have your Say · Initiative 16112

Amending CIRs (EU) 2024/2977, 2024/2979, 2024/2980 and 2024/2982 – Implementing Regulation (EU) 2026/1731 · Statement of October 28, 2026

Core statement: Standards are structure, and GISAD welcomes the maintenance of the wallet standards in an open European standardisation process. But this standardisation also decides two limits: interoperability must not become linkability – a wallet that leaves traces everywhere with every presentation is not proof of identity, but a motion detector. And wherever AI agents present credentials on behalf of a person, the rule of Trusted WEB 4.0 applies: single-step release, signed instance numbers, no standing authorisation. The standard determines whether the wallet organises trust – or profiling.

What the regulation provides

Implementing Regulation (EU) 2026/1731 of 15 July 2026, in force since 11 August 2026, amends the four wallet implementing regulations CIR (EU) 2024/2977, 2024/2979, 2024/2980 and 2024/2982 by updating the references to applicable standards and technical specifications, ensuring that Member States can develop and provide European Digital Identity Wallets in an interoperable manner (Article 5a(23) of the eIDAS Regulation). The consultation (5 February to 5 March 2026, 43 contributions) ran in parallel to the two sister consultations on relying parties and attestations of attributes. What is affected is the technical core on which Member States are building their wallets – a few months before the wallets must be available on 24 December 2026.

Recognition: open standards in European hands

GISAD acknowledges that the standards are developed in the European standardisation process (ETSI TC ESI), published as legal acts and consulted before adoption: transparent, comprehensible, European. This is structure instead of vendor lock-in – no proprietary formats of individual providers, no de facto dependencies on non-European standard setters. The specification of an interface for data deletion requests to relying parties is also a step in the right direction: data minimisation becomes a technical obligation, not a declaration of intent.

Criticism 1: interoperability is not linkability

The amended CIRs require wallet units to validate access certificates at every interaction, check wallet unit attestations and present attributes. Technically correct – but each of these checks is an event that leaves traces with relying parties and in protocols. Anyone using the same wallet at twenty services leaves twenty linkable data records, even if each individual service sees only one attestation. The standard must prevent linkage, not merely minimise individual disclosures: no cross-service correlation identifiers, WAN-anonymous metadata, assignment only internal, following the transaction number model of the EU-D-S. The individual presentation is anonymised – the sum of the presentations must not produce a profile.

Criticism 2: the presenting agent is the escape case

The consultation called for agent-mediated machine-to-machine interactions to be explicitly included in the specifications and interoperability tests: AI agents that request and present attestations on behalf of the user. This is the constellation that Trusted WEB 4.0 resolves with the Finder-KI-Escape-Schalter – and the security signals of recent months show it is not theory: models from OpenAI and Anthropic escaped closed test environments, and the Hugging Face attack of July 2026 happened with reduced guardrails precisely in a testing environment. An AI agent with credential function therefore needs: single-step release for every presentation (no standing authorisation, no background synchronisation), signed instance numbers from the Truststation for every agent instance, measurement outside the manipulable area, and the zero-tolerance rule for instance-identifying markers outside the system.

Criticism 3: the target moves shortly before the start

The technical core of the wallets is being amended five months before the availability deadline. Transparency and consultation are good – but Member States building against the previous reference standard face rework, and interoperability tests must run against a target that no longer moves. Structure protects only if it stands before the operation begins; standard maintenance needs fixed reference points and adequate transition periods.

GISAD’s recommendations

1. Linkage protection as a principle of standardisation: no cross-service correlation identifiers in wallet unit attestations; WAN-anonymous metadata; assignment only internal (transaction number principle of the EU-D-S).
2. Agent-mediated interactions only with single-step release per presentation, signed instance numbers of the Truststation per agent instance and the zero-tolerance rule; no standing authorisations in the specifications.
3. Data deletion requests not as an optional interface, but as a mandatory component of relying party compliance.
4. Fixed reference standards with adequate transition periods; with every standard update, transition periods that do not force Member States into duplicate work.
5. Extend interoperability tests to WAN-anonymous use cases: access and presentation without identification where identity is not legally required – the wallet must be able to do what the border rule of the EU-D-S demands: identity only where identity is needed.

European Digital Identity Framework – validation of qualified/advanced electronic signatures and seals

European Commission · Have your Say · Initiative 16072

Implementing Regulation (EU) 2025/1945 on the validation of qualified and advanced electronic signatures and seals · Statement of October 28, 2026

Core statement: A signature is only worth as much as its validation. GISAD welcomes the harmonisation of validation through open reference standards and decentralised trust lists – it is the right infrastructure for trust. But validation must not become identification: validating a signature confirms its validity and the eligibility it certifies, not the exposure of the person using it. Whoever conflates the two is not building a trust infrastructure, but a surveillance infrastructure.

What the regulation provides

Implementing Regulation (EU) 2025/1945 of 29 September 2025, published in the Official Journal on 30 September 2025 and applicable since 20 October 2025, sets out the reference standards for the validation of qualified electronic signatures and seals (Annex I) and of advanced electronic signatures and seals based on qualified certificates (Annex II): ETSI TS 119 172-4 (validation policy for European qualified signatures and seals using trusted lists) and ETSI TS 119 102-2 (signature validation report), each with specified adaptations. The act is based on Articles 32(3), 32a(3), 40 and 40a of the eIDAS Regulation (EU) No 910/2014. A companion act, Implementing Regulation (EU) 2025/1942, governs the qualified validation services themselves.

Recognition: testing by open rules is structure, not arbitrariness

GISAD expressly acknowledges two design decisions. First: the reference standards are open, documented and comprehensible to everyone – validation becomes a measurable check instead of a black-box verdict. This corresponds to the principle of measurement outside the manipulable area: the validation report (ETSI TS 119 102-2) makes the check itself reviewable. Second: the trust anchors lie in the trusted lists of the Member States – distributed decentrally, not in one central register that could become a single point of failure or a central point of access.

Criticism 1: validation confirms the signature, not the person

The regulation governs the technical validity check. The danger lies in the practice that follows: when recipients automatically turn every validated signature into an identification event, and validation processes are logged, a movement profile emerges – when, where and with which system a signature was checked and used. The signature confirms eligibility; it does not have to identify the person. The EU-D-S separates precisely this with its border rule upon entering Trusted WEB 4.0: identity only where identity is needed. The signed instance number issued by the Truststation is the model – it confirms eligibility without identifying the user, and transaction numbers can only be assigned internally. Pseudonymous qualified certificates, which eIDAS provides for, must be fully supported in validation policies as a normal case, not treated as an exception.

Criticism 2: the validator must not become the gatekeeper

Validation must remain possible everywhere and by anyone who applies the open standards – not de facto only through qualified validation services and vendor-specific verification tools. If checking is tied to a few service providers, a new gatekeeper level emerges: whoever checks decides what counts. The chain must not read: trust is generated by the signature, the signature is checked by the service provider, the service provider decides the benefit. Trust must lie with the structure – open standards, comprehensible reports, decentralised trust anchors.

GISAD’s recommendations

1. Validation policies must strictly separate signature validity from signer identity; pseudonymous qualified certificates as an equivalent normal case, not as an exception.
2. Keep validation logs data-minimal: no central retention of checking processes that create movement profiles; assignment only internal, following the transaction number model of the EU-D-S.
3. Guarantee independent validation by anyone using open standards – the validation report (ETSI TS 119 102-2) as the standard of comprehensibility; no de facto dependence on individual service providers or verification tools.
4. Keep trust anchors decentralised in the trusted lists of the Member States – no central EU trust register as a single point of failure.
5. Adopt the border rule of the EU-D-S as validation practice: the check confirms eligibility (signed by the Truststation) without identifying the user – WAN anonymity up to the point where identity is legally required.

EU copyright law for blind and visually impaired people – evaluation of the Marrakesh Directive and Regulation

European Commission · Have your Say · Initiative 13615

Evaluation of the Marrakesh Directive (EU) 2017/1564 and Marrakesh Regulation (EU) 2017/1563 · Statement of October 28, 2026

Core statement: An evaluation report that ends with „no amendments warranted“ measures administration, not participation. The Marrakesh Directive and Regulation are the right instrument – but as long as the overwhelming majority of published books remains inaccessible to blind and visually impaired people, the objective has not been reached. Participation is not created through exceptions to copyright, but through structure: accessibility as the default state of digital publishing, access without repeated proof of disability, and availability that is measured publicly.

What the report finds

The Commission has evaluated the Marrakesh Directive (EU) 2017/1564 and the Marrakesh Regulation (EU) 2017/1563, through which the EU implements the 2013 WIPO Marrakesh Treaty. The evaluation, required by Article 10(1) of the Directive and Article 7 of the Regulation, also assessed the impact of possible compensation schemes on the availability of accessible format copies and their cross-border exchange. The result, published on 12 December 2025 as report COM(2025)749 with the accompanying staff working document SWD(2025)406: the instruments matched expectations, improved legal certainty and the availability of accessible format copies, including across borders – and no amendments are considered warranted at this stage. The call for evidence ran from 13 April to 11 May 2023 and received 34 contributions, among them the DAISY Consortium and national libraries for the blind.

Recognition: cross-border exchange without a central platform

GISAD acknowledges what the report records as a success and what is structurally remarkable: the cross-border exchange of accessible format copies works through authorised entities in the Member States – decentralised, without a central EU platform, without a single point of failure. This is precisely the principle the EU-D-S is built on: many authorised entities complementing each other, instead of one body deciding for all. The participatory character of the instrument is right: Marrakesh recognises access to knowledge as participation, not as welfare.

Criticism: the exception is measured, not the condition

The yardstick of an evaluation determines its outcome. The Commission measures legal certainty, administrative procedures and cross-border handling – but not the actual availability of accessible titles relative to the totality of publications. According to estimates by the European Blind Union, around 30 million people in Europe are blind or partially sighted, and 95% of all published books are never converted into accessible formats; a study across six European countries finds less than 7% of published works accessible. An evaluation that concludes „no amendments warranted“ against this background confuses smooth administration with a fulfilled purpose. What is not measured publicly is not improved – the availability measurement therefore belongs in the report, independent and visible for every Member State: measurement outside the manipulable area.

Structure instead of welfare: access without an ID requirement

Using the exception runs through authorised entities and proof of eligibility. That is better than no access – but it turns participation into an administrative relationship with an ID requirement: whoever wants to read must prove their disability again and again. The EU-D-S has a simpler rule for this: identity only where identity is needed. A one-time, Truststation-signed credential – followed by WAN-anonymous access without profiling, without repeated verification of disability, without stigmatisation. The eligibility is checked, the person is not registered.

Compensation schemes: those who end the book famine do not pay those who caused it

The evaluation also examines possible compensation schemes. GISAD is clear on this: compensation models that burden non-profit libraries and organisations turn responsibility upside down. The Treaty became necessary because publishers did not provide accessible formats at the same time and at the same price. Those who close this gap must not be charged extra – that would be a participation levy on precisely those who create participation. The financing question belongs where the EU-D-S locates it: in participation payments and structure, not in levies on non-profit work.

GISAD’s recommendations

1. Public availability measurement as a core indicator: annual, independently collected figures on accessible titles per Member State, relative to the totality of publications – measurement outside the manipulable area.
2. Accessibility as the default state („born accessible“): accessible formats as a requirement of digital publishing itself – structure instead of retrofitted exception; the exception remains what it should be: a last line of defence, not the rule.
3. No compensation schemes at the expense of authorised entities; resolve the cost question through participation mechanisms instead of levies on non-profit work.
4. Proof of eligibility one-time and signed (Truststation), then WAN-anonymous access without profiling – identity only where identity is needed.
5. Keep the decentralised exchange through authorised entities as the guiding model – no central EU platform that becomes a single point of failure; linguistic diversity including minority languages as the yardstick (getmySense thinks in 2,500 written languages – participation applies to every language community, and to the elderly, for whom large print and read-aloud functions are participation).

European Commission · Have your Say · Initiative 15572

European Digital Identity Wallets –
user onboarding

Implementing Regulation (EU) 2026/798  ·  eIDAS 2.0  ·  Statement of 28 October 2026

European Commission · Have your Say · Initiative 15514

Review of the
Digital Decade policy programme

Digital Decade 2030  ·  Statement of 28 October 2026