AIPS Digital Product Passport (DPP) Overview
Preamble
The concept of the Digital Product Passport (DPP) originates from the European Union’s Circular Economy Action Plan (CEAP) and the Ecodesign for Sustainable Products Regulation (ESPR).
Under these initiatives, the EU aims to make product-related information — such as composition, origin, environmental impact, and lifecycle data — digitally available and verifiable.
The Ecodesign for Sustainable Products Regulation (ESPR) (proposal: COM/2022/142 final) requires most physical products placed on the EU market to include a DPP that supports traceability, circularity, and compliance.
At present, the regulation applies primarily to physical goods (e.g., textiles, batteries, electronics) and does not yet mandate coverage for non-physical products such as data, AI, or digital services.
Nevertheless, the DPP concept has inspired the creation of digital analogues for non-physical domains — such as Data Products and AI Products — to promote trust, accountability, and transparency.
The AIPS DPP initiative extends these principles to AI systems by combining the BPS DPP Core with AI-specific sections related to model cards, evaluation results, training data provenance, and AI risk information.
Purpose
The AIPS DPP defines a domain-specific profile of the BPS DPP Core for AI Products.
It enables transparent and verifiable disclosure of critical metadata, evaluation metrics, and provenance details associated with AI models — whether for internal governance, regulators, or external consumers.
Scope
| Aspect | Description |
|---|---|
| Domain | AI Products and related model artifacts (e.g., models, datasets, pipelines). |
| Foundation | Composes directly with BPS DPP Core 0.1 for headers, integrity, and structure. |
| Profile IRI | https://kivanura.org/spec/aips/dpp/0.1/ |
| Core Composition | AIPS DPP reuses BPS DPP’s core sections (schema, lineage, provenance, quality, policy, kg) and adds AI-specific inline/ByRef sections. |
| Artifacts | Context, SHACL, Schema, TTL, OpenAPI definitions under /spec/aips/dpp/0.1/. |
| Usage | For each AI Product, producers publish a Lite and Full DPP accessible via API or QR code in marketplaces. |
Relationship to BPS DPP Core
The AIPS DPP builds on the BPS DPP Core as follows:
| Function | Provided by BPS DPP Core | Extended by AIPS DPP |
|---|---|---|
| Product identity and issuer | ✅ Yes | — |
| Signatures and integrity hashes | ✅ Yes | — |
| Schema, lineage, provenance, quality, policy, KG sections | ✅ Yes | Enhanced with AI context |
| Domain-specific sections | — | ✅ Adds model card, evaluation, training data, risk |
| Validation | ✅ SHACL rules | ✅ Additional SHACL constraints for AI |
| Namespace | https://kivanura.org/spec/bps/dpp/0.1/ | https://kivanura.org/spec/aips/dpp/0.1/ |
Design Principles
- Composition over Duplication — AIPS DPP does not redefine core DPP properties; it extends them through its own namespace.
- Layered Validation — First validate against
bps-dpp-core-0.1.shacl.ttl, then applyaips-dpp-0.1.shacl.ttl. - Lite and Full Views — Both versions share the same schema; Full includes additional inline and ByRef data.
- Verifiability — Every AIPS DPP must include a
contentHashandsignaturefield compliant with BPS integrity requirements. - Semantic Reuse — Reuses terms from W3C PROV-O, DCAT, and Schema.org where appropriate.
Example Use Case
A financial institution releases a Fraud Detection Model v2.1 as an AI Product.
Before onboarding, external consumers (e.g., business partners, regulators) can access the model’s Lite DPP via a QR code, showing:
- Model name, framework, and version
- Intended use and license
- Summary evaluation metrics
- Policy findings and compliance status
Authorized users can then access the Full DPP, which includes:
- Provenance and lineage metadata
- Model card and evaluation datasets
- Risk and mitigation details
- Digital signature for verification
Summary
The AIPS DPP:
- Extends the BPS DPP Core to AI systems.
- Provides a standardized, verifiable passport for AI Products.
- Bridges technical transparency (evaluation, provenance) with governance (policy, accountability).
- Lays the foundation for trustworthy AI Product ecosystems across enterprise and regulatory domains.