Skip to content

GS1 Digital Link And Content Addressed Product Information

Status And Scope

Design and analysis note, 10 October 2026. Describes the implemented OpenQR artifact-link model and a proposed structured-product-record extension. It does not allocate GS1 identifiers, define a new event kind, or claim GS1 certification.

The policy brief provides the public-policy overview. The OpenQR implementation note documents exact FastAPI behaviour and validation boundaries.

Complementary Responsibilities

For scanning existing GS1 element strings without a digest, see the GS1 Barcode Acquisition And Anchor Discovery specification. It defines GTIN relay discovery followed by local lot and expiry matching, including separate anchors for different lot/expiry combinations.

GS1 Digital Link provides product identification in a web-compatible form. A compatible checkout application extracts the identifiers; a consumer browser opens the web resource. OpenETR adds signed evidence concerning an exact digital artifact rather than replacing the retail identification system.

Concern Mechanism Limit
Product identification GTIN and optional lot/serial Requires appropriate allocation and usage
Convenient navigation HTTPS resolver domain Availability and operation depend on that service
Artifact byte identity SHA-256 digest Does not guarantee retrieval or truth
Attribution of an association Signed OpenETR anchor Identifies a signing key, not automatically an authorized producer
Presentation Browser, OpenQR, or another reader Codec and document support vary
Recognition and consequence Integrator's accepted evidence and rules Not supplied by the barcode alone

The distinction between symbology and payload matters: QR is the visual carrier; GS1 Digital Link is the URL convention. The chosen approach uses an ordinary QR generator, not FNC1-based GS1 QR mode. Scanner readiness and printed-symbol requirements still need validation. See GS1 URI syntax and retail guidance.

Rulebooks And Semantic Interoperability

Semantic interoperability is a responsibility of domain rulebooks, not an additional universal product model for OpenETR. A rulebook may incorporate GS1, laboratory, product-passport, or other standards by reference. It should identify the applicable versions, vocabularies, units, field mappings, and interpretation rules needed for its purpose.

Keep the following evaluations distinct:

Evaluation Question
Byte integrity Are these the exact bytes identified by the digest?
Signed attribution Does this event verify under the stated signing key?
Semantic interpretation What do the fields and assertions mean under the domain rules?
Recognition Is this issuer or evidence accepted for the stated purpose?
Consequential-state evaluation What state follows from the available evidence under the applicable rules?

OpenETR defines its core evidence conventions. Domain adapters and rulebooks add domain-specific meaning, safeguards, and evaluation requirements. A digest match must not be reported as schema conformity, nor a signature match as regulatory approval. Missing schemas, ambiguous extraction, or unknown units should be reported as unresolved domain checks rather than silently treated as success.

Shared evidence does not require an identical rulebook; interoperable interpretation requires sufficient agreement on meaning. Two verifiers may agree on a certificate's measurements but reach different acceptance decisions for different purposes. Comparisons should expose the evidence scope and rulebook version, including incorporated semantic standards. Local rules cannot waive applicable legal obligations.

See also the DPP analysis.

URL And Digest Convention

Illustrative template:

https://example.com/01/{gtin}/10/{lot}/21/{serial}?d={digest}

Generated links use d, saving five ASCII bytes. The descriptive digest alias remains supported for existing links and authors who prefer it. Readers accept exactly one occurrence of either name, never both, even if their values match. Both accept the same digest encodings; generation and redirects use d.

Lot and serial segments are optional. d (alias digest) is an OpenQR application extension, not a newly defined GS1 Application Identifier. Its generated value is:

base64url_without_padding(SHA-256(exact_artifact_bytes))

The 32-byte digest becomes 43 characters. OpenQR also accepts lowercase hex on input and normalizes to hex for the o event tag and relay filter. No reserialization, transcoding, or normalization is applied to the artifact before hashing.

The digest is not the GS1 serial: AI 21 is limited to 20 characters, and a document can legitimately describe many product instances. Product identity and content version identity must not be conflated.

This is a custom direct-resolution extension under the OpenETR QR Resolver Profile, not its query-free Standard Web Resolver Profile. The GS1 values identify the associated product, not another OpenETR record. OpenQR currently supports only a bounded subset of Digital Link syntax and is not a full GS1-conformant resolver.

Implemented Binding Through The Anchor

For an uploaded video, datasheet, or other artifact, the anchor signs the hexadecimal digest in o together with g (GTIN), optional gs1_lot, and optional gs1_serial. The event is kind 1415 with action issue.

Exact artifact bytes -> artifact digest
Signed anchor -> artifact digest + GTIN + optional lot/serial
Digital Link -> same artifact digest + product identifiers

The artifact need not contain the identifiers. Their association with it is committed by the signed event, not necessarily by the artifact hash itself. Verification retrieves the bytes, checks the digest, verifies the anchor, and compares the product fields. Putting a digest beside identifiers in a URL without checking signed evidence would not establish their association.

The wire format supplies event identity and signature rules. The application tags are structured publisher assertions, not proof that GS1 assigned the identifier to the signer. OpenQR displays mismatches without discarding independently verifiable artifacts or other candidate anchors.

Proposed Binding Within A Product Record

A stronger content-level commitment can be obtained by making a structured product record the artifact itself. An illustrative record could contain:

{
  "profile": "example-product-record-draft",
  "gtin": "09520123456788",
  "lot": "LOT456",
  "serial": "ABC123",
  "resources": [
    {"role": "product-video", "sha256": "<full video digest>", "media_type": "video/mp4"},
    {"role": "datasheet", "sha256": "<full PDF digest>", "media_type": "application/pdf"}
  ]
}

This is a conceptual example, not an implemented schema. The GTIN is documentation test data. A versioned schema, parsing rules, resource limits, and conformance tests must be agreed before deployment.

Hash the exact finalized product-record bytes. Publish an ordinary signed anchor whose o references that hash. The QR then carries the product-record digest, not the video digest. This avoids making the record contain its own digest or hashing a self-referencing signed anchor. The anchor signature remains separate from the artifact hash.

Verification would:

  1. Extract the expected record digest and product identifiers from the link.
  2. Retrieve the exact product-record bytes and verify their hash before parsing.
  3. Parse under the declared schema, rejecting ambiguous fields such as duplicate JSON keys, and compare the product identifiers with the URL.
  4. Verify the anchor's event ID and signature and its reference to the record digest.
  5. Apply issuer-recognition rules independently of signature correctness.
  6. Retrieve referenced media as needed and verify each resource's own digest.

With this design, altering an identifier inside the record changes the digest; altering only the URL's identifier fails the comparison. No separate signature per identifier is needed. If canonical serialization is introduced later, the profile must specify it; verifiers must not silently reserialize exact-byte artifacts. OpenQR does not yet parse this proposed product-record schema.

Resolution Without The Printed Domain

An aware reader can recover the digest from the scanned URL without making a request to its host. It can query its chosen relays by #o, inspect signed anchors and safe storage hints, and obtain the same bytes from another Blossom server, an HTTPS mirror, a local cache, or an archive. A different website can present the material without changing its identity.

This independence has prerequisites: the reader understands the extension, has usable discovery sources, can obtain a copy, and can evaluate the relevant signatures and recognition rules. A digest is not a global discovery service. An ordinary phone camera normally follows the URL and does not perform this alternative resolution automatically. Offline verification needs previously obtained artifact bytes, evidence, and any required recognition material.

OpenQR implements configurable relay and Blossom pools, not arbitrary mirror discovery. Additional reader integrations are an architectural possibility, not a claim that every scanner or resolver already supports them.

Independent Regulatory Acquisition And Certificate Verification

A regulator can implement a purpose-built acquisition application while consumers continue using ordinary mobile QR cameras. The consumer camera opens the HTTPS URL; the regulatory reader parses the scanned text locally and need not contact that URL's domain. GTIN, lot, serial, and digest become separate inputs to the regulator's verification policy rather than an instruction to trust a website.

For a Certificate of Analysis, the proposed workflow is:

  1. Acquire the barcode payload and validate its identifier and digest syntax.
  2. Obtain the original certificate bytes from any permitted source, including a file already held by the regulator. Do not hash a screenshot, printed copy, extracted text, or regenerated PDF as if it were the original digital file.
  3. Compute SHA-256 over those exact bytes and compare the normalized result with the barcode digest. A mismatch fails artifact correspondence regardless of whether the filenames or visible pages look alike.
  4. Extract the certificate's GTIN, lot, and serial under a declared document schema or review procedure. Compare each required field with the barcode, preserving meaningful leading zeros and applying only defined normalization.
  5. Retrieve relevant OpenETR anchors and associated evidence by digest and any applicable event references. Verify event IDs, signatures, and artifact links.
  6. Validate any embedded or separately referenced signatures and seals under their applicable formats, then apply issuer-recognition and authority rules.
  7. Evaluate any required correction, withdrawal, validity-period, or supersession evidence under the regulator's rules, reporting retrieval scope and limitations.

Two assertions must be distinguished: these are the exact committed bytes and these bytes contain the matching identifiers in the relevant fields. The second requires interpreting the document. A hash comparison alone cannot perform it. Documents with multiple products, contradictory identifiers, missing fields, or ambiguous OCR must not pass by a loose substring match. Structured records can support deterministic field checks; ordinary PDF certificates need a specified extraction profile or explicit human confirmation. Batch-level certificates must not be presented as containing an item serial they do not state.

If the barcode digest identifies a product manifest rather than the certificate itself, first verify and parse the manifest, then verify the certificate against its referenced digest. The certificate digest must not be compared directly with the manifest digest. An indirect reference must be made explicit in the result.

When the certificate itself is the hashed artifact, matching its bytes and fields creates a cryptographically verifiable binding to the acquired barcode payload, subject to the hash's security assumptions and correct parsing. It is not an incontrovertible physical or legal fact: a barcode can be copied or replaced, an issuer can make false assertions, and a valid certificate can later be withdrawn.

OpenETR signatures and any applicable associated evidence can be discovered using the wire-format conventions. Retrieval is not equivalent to recognizing a seal. The electronic seals design note distinguishes proposed sealing profiles from implemented core signature verification. An integration must specify how a seal or external signature is referenced, what bytes it covers, what validation material is needed, and whose authority is accepted. Finalized embedded signatures or seals are part of the exact bytes being hashed; adding one later changes the artifact digest.

Current OpenQR supports digest lookup, byte verification, anchor-signature checks, and comparison with signed GS1 tags. Automated certificate field extraction, external signature/seal validation, and this regulator-specific evaluation workflow are integration work, not existing OpenQR functionality. A future regulatory pilot should test independent acquisition from a held file as well as retrieval with the printed domain unavailable.

Visual Confirmation And Optional Identity Applications

A product record may include a photograph to support visual inspection. The photograph must be embedded in the exact hashed artifact or referenced by digest from a verified manifest. A mutable image URL alone does not bind the displayed image to the record. For a referenced image, verify its bytes against its own digest before rendering; preserve the distinction between the manifest digest and the image digest.

The record should identify whether the image depicts a product model, a batch, or the particular serialized item, and state any asserted capture context. An inspector can compare packaging, markings, and visible characteristics with the verified image. Report artifact integrity, identifier correspondence, issuer recognition, and visual assessment separately. A matching photograph does not prove provenance, unseen properties, or that the label has not been copied.

As an optional extension outside this product scenario, a passport- or driver's-licence-related record could contain a portrait or biometric template. An officer could manually compare the portrait with its presenter; compatible software could assist portrait comparison or compare a newly acquired biometric sample with the verified template. A template is not necessarily a viewable image and requires a declared format and compatible comparison process.

Such an integration would need to define:

  • The exact artifact bytes, image or template references, and signature coverage.
  • Recognition of the issuing authority and checks for expiry, revocation, or supersession under the applicable rules.
  • Capture and comparison procedures, uncertainty reporting, human review, and safeguards against replay or other presentation attacks where required.
  • Access authorization, encryption where appropriate, retention limits, and controls on disclosure and cross-context correlation.

Do not put personal images or biometric templates into public relay events or publicly retrievable storage by default. A digest does not conceal accessible content, establish consent, or anonymize a person. Protected retrieval can still allow an authorized verifier to check the bytes independently. A digest match verifies reference material, while a biometric or visual match is a distinct assessment with its own limitations; neither alone establishes legal identity.

These examples illustrate an extensible verification architecture. They do not define a new identity-document standard, require GS1 identifiers for people, or claim that OpenQR currently implements biometric comparison or identity checks.

Physical Trade And Tamper-Evident Attachment

Independent document verification can reduce uncertainty in physical trade and help participants concentrate inspection on correspondence between the verified record and the goods presented. It does not reduce every trade risk to a physical match: issuer authority, record currency, quality claims, and transaction rules remain separate considerations.

A tamper-resistant sticker bearing the barcode is one possible attachment mechanism. An integration should specify the item or packaging to which it is attached, who applies it, and how inspectors assess removal or replacement. Destructible and tamper-evident features can reveal interference; they should not be described as tamper-proof or as making the printed payload uncopyable.

A physical-trade pilot should test:

  • Controlled label application and correspondence with the item's serial number.
  • Inspection of label condition alongside verified photographs and identifiers.
  • Removal, copying, substitution, and opening or replacement of labelled packaging.
  • Recording handover observations, including who inspected what and when, where the integration requires attributable inspection evidence.
  • Legitimate repackaging or relabelling, with explicit rules for preserving the relationship to earlier records rather than silently reusing an old label.

A package-level seal does not by itself prove its contents or establish title. Report digital verification and physical inspection outcomes separately. No new OpenETR action or transfer ruleset is introduced by this attachment convention.

Property /{campaign}/{digest} GS1 Digital Link
Digest location Path d query parameter (digest also accepted)
Product identifiers Not in URL GTIN, optional lot/serial
Dispatch Campaign passed through; currently no campaign rules Default etr handler
Evidence lookup Kind 1415 plus #o Same
Indirection database Not needed Not needed
Product-association check Not requested by URL Compared with signed GS1 tags
Retail meaning No standard GTIN structure Interpretable by compatible GS1 systems

Neither current route is an opaque campaign-token mapping. Both carry the digest directly and converge on the same lookup and rendering functions. Incoming GS1 fields are validated before the lookup; they do not become relay filter keys.

Operational And Policy Boundaries

Changing file bytes produces a new digest. The printed code remains committed to the old version. A later correction, withdrawal, or supersession therefore needs separate evidence and explicit rules; anchor discovery alone is not a current-state determination. Selective or incomplete relay responses can omit relevant notices.

Replication improves availability but is not guaranteed by signatures. Access restrictions remain necessary for confidential information; public QR codes and public storage should not expose private product or customer data inadvertently.

A copied label yields the same valid references as the original. Label replacement can substitute an entirely different digest and signing key. Recognition, physical anti-counterfeit measures, and any ownership or redemption rules remain separate. An ordinary point-of-sale scan does not constitute an OpenETR control transfer.

For small labels, the additional URL data increases symbol density. Test the actual payload, print process, margins, error correction, bottle curvature, and scanner population. A scratch-off label intended for post-purchase access may need to be separate from the visible checkout code. See the production design note.

Pilot Acceptance Criteria

  1. Verify GS1 identifier allocation and test checkout behaviour with the retailer.
  2. Confirm native phone navigation and artifact rendering for intended formats.
  3. Verify signed product associations and reject altered or duplicate identifiers.
  4. Retrieve and verify the same bytes with the printed domain unavailable.
  5. Test missing copies, unavailable relays, conflicting assertions, and replaced labels.
  6. Define version and later-notice handling before making current-status claims.

The intended outcome is provider-independent verification of product information, not an assertion that cryptography alone establishes product truth or legal effect.