Requirements for the AI BOM that a supplier providing AI models or systems to your organization must submit. Defines the standard data format, the information that must be included, identifier rules, and the requirements for licensing, provenance, and sensitivity.
Guidelines for the AI BOM that in-house development teams must produce when building an AI model or system. Drawing on the information accessibility available at the point of production, this sets out the required and recommended fields to fill and how to record integrity and provenance.
A checklist for vetting transparency and risk, on the basis of the AI BOM, when bringing in and using an external AI model or dataset. Checks identification, licensing, data lawfulness, and security risk step by step.
Surveys seven tool categories through their official repositories and documentation to lay out what to reuse, extend, or build new; the build order; the policy schema for codifying the matrix; and the Dependency-Track integration architecture.
Weighs the 50 elements of the G7 “SBOM for AI — Minimum Elements” against authoritative standards — SPDX 3.0.1, CycloneDX 1.6, NTIA 2021, OpenChain AI V1 — and regulatory grounds including the CRA, the AI Act, and FDA guidance, to determine which AI BOM fields are required and which are optional. Part of a five-part series that also applies the same matrix to production, ingestion, and supplier contexts and covers toolset strategy.
Covers SBOM generation, management, and scanning tools, and where automation ends and human responsibility begins.