◎ OS PUB Apache 2.0 ← All specifications

P133 — AIEP — Distributed Evidence Index Protocol

Publication Date: 2026-03-15 Status: Open Source Prior Art Disclosure Licence: Apache License 2.0 Author/Organisation: Phatfella Ltd Schema: AIEP_OS_SPEC_TEMPLATE v1.0.1 — https://aiep.dev/schemas/aiep-os-spec-template/v1.0.1

Framework Context

[0001] This disclosure operates within an Architected Instruction and Evidence Protocol (AIEP) environment as defined in United Kingdom patent application number GB2519711.2, filed 20 November 2025, and GB2519803.7, filed 20 November 2025, the entire contents of which are incorporated herein by reference.

[0002] The present disclosure defines an open protocol for distributing, indexing, and discovering verifiable evidence artefacts across heterogeneous AIEP-governed systems and the open web, building on the evidence substrate defined in the AIEP core specifications.

Field of the Disclosure

[0003] This disclosure relates to distributed indexing protocols for verifiable evidence artefacts in governed artificial intelligence reasoning systems.

[0004] More particularly, the disclosure concerns a protocol under which evidence nodes containing structured metadata and cryptographic hash references to source material are published, discovered, and retrieved by AIEP-governed reasoning systems without centralising the underlying knowledge, transforming the open web into a distributed AI memory substrate.

Background

[0005] AIEP-governed reasoning systems require evidence artefacts to be verifiable, retrievable on demand, and stable across time. Existing web content is neither structured for machine-verifiable reasoning nor indexed in a form that allows AI systems to discover relevant evidence with jurisdiction, domain, and provenance metadata attached.

[0006] Centralising evidence storage within individual AI systems defeats the purpose of evidence-grounded reasoning: any single evidence store becomes a single point of failure, a potential vector for undetected evidence manipulation, and a constraint on the scale of the evidence base that can be maintained.

[0007] No existing protocol defines a normative evidence node schema, a distributed index entry format, or a discovery mechanism that allows multiple AIEP-governed systems to share and discover the same evidence artefacts without sharing model state or reasoning history.

Summary of the Disclosure

[0008] An EvidenceNode contains structured metadata describing a source artefact: source_url; mirror_url (if a structured mirror has been created); content_hash computed as SHA-256 over the retrieved content body; retrieval_timestamp in ISO 8601 format; source_type from a schema-defined enumeration; jurisdiction as an ISO 3166-1 alpha-2 or ISO 3166-2 code; and classification_tags as a schema-defined taxonomy array.

[0009] An EvidenceIndexEntry extends the EvidenceNode with fields required for discovery: evidence_id derived as the SHA-256 hash of the canonical serialisation of source_url, content_hash, and retrieval_timestamp; topic_tags; language as a BCP 47 tag; and relevance_score as a floating-point value in [0,1] computed at indexing time by the submitting node.

[0010] Evidence Discovery proceeds by: (a) classifying the incoming query against the schema-defined topic taxonomy; (b) querying the index by topic tags, jurisdiction, and recency constraints; (c) retrieving and verifying content hash integrity for each candidate artefact; (d) ranking artefacts by relevance score; and (e) admitting artefacts to the evidence rail subject to the governance admissibility gate.

[0011] Index Distribution is achieved by allowing any AIEP hub, organisation, academic institution, or sovereign node to host an index partition. Index partitions publish their entry set at a canonical well-known endpoint following the format defined in P61. A consuming node may aggregate multiple index partitions without duplicating evidence storage.

[0012] Verification of a retrieved artefact is performed by: recomputing SHA-256 over the retrieved content body and comparing against the stored content_hash; confirming that retrieval_timestamp falls within the defined staleness window; and confirming that jurisdiction and classification_tags are consistent with the schema-defined vocabulary.

[0013] Artefacts whose content hash does not match the stored hash, or whose retrieval timestamp exceeds the staleness window, are rejected and a NegativeProofRecord is issued identifying the artefact and rejection reason.

Technical Effect

[0014] The invention achieves the following technical effects:

[0015] Content-hash-bound evidence identifiers ensure that any tampering with source material after indexing is detectable by any node that retrieves and verifies the artefact, providing tamper evidence without requiring a centralised trust authority.

[0016] Distributed index partition hosting eliminates single-point-of-failure evidence stores: the failure or compromise of one index partition does not prevent discovery of artefacts indexed by other partitions.

[0017] Schema-defined jurisdiction and classification metadata enables jurisdiction-aware evidence retrieval, allowing reasoning systems operating in multiple regulatory contexts to filter evidence by applicable jurisdiction without manual curation.

[0018] Publishing the evidence node schema and index entry format as open prior art prevents competitors from claiming proprietary rights over the distributable evidence indexing layer, ensuring that any AIEP-governed system can participate in the evidence index ecosystem.

Claims

A distributed evidence indexing protocol for governed AI reasoning systems, the protocol comprising: an EvidenceNode schema including a content hash computed over retrieved source content, a retrieval timestamp, and jurisdiction and classification metadata; an EvidenceIndexEntry extending the EvidenceNode with a deterministically derived evidence identifier, topic tags, and a relevance score; and an evidence discovery procedure classifying a query against the topic taxonomy, querying the index, verifying content hash integrity on retrieval, and admitting artefacts through a governance admissibility gate.
The protocol of claim 1, wherein the evidence identifier is derived as the SHA-256 hash of the canonical serialisation of the source URL, content hash, and retrieval timestamp, providing a deterministic, collision-resistant identifier that is reproducible by any node given the same inputs.
The protocol of claim 1, wherein artefacts whose retrieved content body does not produce a SHA-256 hash matching the stored content hash are rejected and a NegativeProofRecord identifying the artefact and the mismatch reason is issued by the rejecting node.
The protocol of claim 1, wherein index partitions are hosted by independent nodes and publish their entry sets at a well-known endpoint following the format defined in the AIEP Well-Known Manifest Specification, enabling consuming nodes to aggregate multiple partitions without duplicating evidence storage.
The protocol of claim 1, wherein jurisdiction is expressed as an ISO 3166-1 alpha-2 or ISO 3166-2 code, and wherein evidence discovery may be filtered by jurisdiction code, enabling reasoning systems operating under jurisdiction-specific regulatory constraints to retrieve only artefacts applicable to the relevant jurisdiction.

Brief Description of the Drawing

FIG. 1 — Evidence Discovery Flow: query classification → index search (multi-partition) → artefact retrieval → content hash verification → governance admissibility gate → evidence rail admission.

FIG. 2 — EvidenceIndexEntry schema fields and derivation relationships.

FIG. 3 — Distributed index partition hosting topology showing AIEP hub, organisational, and sovereign node partitions with well-known endpoint publication.

Abstract

A distributed evidence indexing protocol defines an EvidenceNode schema — including a content hash, retrieval timestamp, jurisdiction, and classification metadata — and an EvidenceIndexEntry extending the node with a deterministically derived identifier and relevance score. Evidence discovery proceeds by query classification, multi-partition index search, content hash verification, and governance admissibility gating. Index partitions are hosted by independent AIEP hubs, organisations, and sovereign nodes, each publishing entries at a well-known endpoint. Artefacts whose content hash fails verification produce a NegativeProofRecord. The protocol transforms the open web into a distributed evidence substrate for governed AI reasoning systems without centralising evidence storage.

Licence

Apache License 2.0 — https://www.apache.org/licenses/LICENSE-2.0