AI SBOM and Regulation
An AI SBOM is a machine-readable inventory of a model, its training datasets, and their provenance and licenses. Where a software SBOM carries package dependencies, an AI SBOM carries the layers a software SBOM cannot express: model weights and training data.
It is built from what the model card says, so a thin model card yields a thin AI SBOM.
Why produce one
It serves two purposes.
For the publisher it is a self-check on how complete the documentation is. Seeing what is missing as a list means you can close the gaps before release.
For the recipient it is evidence for a review. Just as SK Telecom asks suppliers for an SBOM, other organizations have begun asking the same of AI models.
The G7 minimum elements
In May 2026, CISA and G7 partners jointly published “Software Bill of Materials for AI — Minimum Elements”, led by Germany’s BSI and Italy’s ACN. It defines 50 minimum elements in seven clusters that a model’s inventory should carry. It is a non-binding recommendation, not a regulation.
| Cluster | Elements | Needing human judgement | Content |
|---|---|---|---|
| Metadata | 10 | 0 | Who produced the inventory, when, with which tool |
| System-level properties | 9 | 4 | System name, application area, data flow |
| Models | 13 | 0 | Model identifier, license, integrity, training properties |
| Dataset properties | 10 | 5 | Dataset provenance, statistics, sensitivity, license |
| Infrastructure | 2 | 0 | Software dependencies and hardware |
| Security properties | 4 | 3 | Security controls, policy, vulnerability handling |
| Key performance indicators | 2 | 1 | Security metrics and operational performance |
Thirteen of the 50 have no automated source. Things like the intended application area or the sensitivity of the training data cannot be proven by any model card field, so a person has to supply them.
How it relates to regulation
The G7 elements are advisory, but they overlap substantially with the technical documentation actual regulations require.
The EU AI Act applies its high-risk and transparency obligations from 2 August 2026. The technical documentation required by Article 11 and Annex IV covers a system’s purpose and architecture, the provenance and processing of its training data, and its performance and limits. The system-level, model and dataset clusters correspond to these.
Korea’s AI Framework Act took effect on 22 January 2026. Article 31 (transparency), Article 32 (safety), Articles 33 and 34 (high-impact AI) and Article 35 (impact assessment) are the relevant provisions. Its detailed documentation requirements sit in the enforcement decree, so the correspondence is coarser than for the EU. Article 32 applies only to systems trained above a compute threshold set by that decree, so a linked element points at the subject of the duty rather than establishing that the duty applies.
An AI SBOM and its regulatory mapping do not certify or determine compliance with any regulation. They make documentation gaps visible so a person can prepare. Interpreting them against a specific system’s legal obligations is a person’s job; when in doubt, consult Legal and the OSRB.
Producing one
Give BomLens a model id and it reads the model card, builds a CycloneDX AI SBOM, and reports G7 element coverage alongside the regulatory mapping. It fetches model-card metadata only and does not download the weights.
./scripts/scan-sbom.sh --project my-llm --version 1.0.0 \
--model "my-org/my-llm" --generate-only
Scanning an AI model needs a separate image, and it can also be run from a web interface. Setup, usage and how to read the reports are covered in the BomLens AI model guide; this page does not repeat them.
A private repository needs a token with read access. For checking a model before release, see Releasing an AI model.
You get:
- an AI SBOM covering the model and its datasets
- per-element coverage, and how to supply what is missing
- the mapping to the EU AI Act and Korea’s AI Framework Act
- components whose license needs human review
Reading the result
An overall pass does not mean every G7 element is filled. G7 elements are all advisory and do not move the verdict. Read the covered count and the gap count separately.
Gaps come in two kinds: those you close by writing in the model card, and those a person has to judge. Start with the first kind.
Related pages
- Model card — the input to an AI SBOM
- Pre-release checklist — what to confirm before release
- What is an SBOM — SBOMs in general
- BomLens AI model guide — setup, usage, reading the reports
- BomLens — the tool, and its supplier-facing usage
References
- Software Bill of Materials for AI — Minimum Elements (G7, May 2026)
- Regulation (EU) 2024/1689 (AI Act)
- AI Framework Act (Korea)
Contact
For questions about producing or reading an AI SBOM, contact the OSRB (opensource@sktelecom.com).
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.