Home / Glossary / AI BOM (AI bill of materials)
Glossary
What is an AI BOM (AI bill of materials)?
An AI BOM (AI bill of materials) is a formal, machine-readable list of the parts that make up an AI system. It records the models, the datasets and the software libraries, and where each one came from. An AI BOM describes one system in depth, and an AI inventory lists all the AI in use across a company.
By the AIBound team · Updated 17 November 2026
Key takeaways
- An AI BOM lists the models, datasets and software inside one AI system, and where each part came from.
- It extends the software bill of materials (SBOM) to cover the parts that are specific to AI.
- An AI BOM describes a system someone built, and an AI inventory shows which AI employees run.
Why an AI BOM matters
An AI system is put together from parts that other people made. The base model may come from a public library. The training data may come from several sources, each with its own licence. When a flaw is found in one part, a team needs to know which of its systems use that part. An AI BOM answers that in a search.
Two open standards cover this. CycloneDX has an ML-BOM that records models, datasets and configurations. SPDX describes an AI system bill of materials that also covers prompts and AI agents. A standard format means a buyer, an auditor or a security tool can read the same file.
How an AI BOM works
An AI BOM is written by the team that builds or supplies the system, and it is kept up to date as the system changes.
- 1 The team lists each part: models, datasets, libraries, prompts and the tools the system calls.
- 2 For each part it records the version, the source, the licence and the supplier.
- 3 The list is saved in a standard format, so other tools can read it.
- 4 The team updates it when a part changes and checks it against new vulnerability reports.
AI BOM vs AI inventory
An AI BOM goes deep on one system. An AI inventory goes wide across the company. The inventory says which AI exists and who runs it. The AI BOM says what a given system is made of. A governance program uses both.
| Question | AI BOM | AI inventory |
|---|---|---|
| What does it list? | The parts inside one AI system | Every AI in use across the company |
| Who writes it? | The team that builds or supplies the system | The security or governance team |
| What does it answer? | What this system is made of, and where each part came from | Which AI employees run, and who runs each one |
Sources
- CycloneDX: Machine Learning Bill of Materials · cyclonedx.org
- SPDX: AI · spdx.dev
Sources checked 17 November 2026
Examples
AI BOM examples
FAQ
Frequently asked questions
What is an AI BOM (AI bill of materials)?
An AI BOM is a formal, machine-readable list of the parts inside an AI system. It names the models, the datasets and the software libraries, with the version, source and licence of each. It lets a team, a buyer or an auditor see what a system is made of.
What is the difference between an AI BOM and an SBOM?
An SBOM, or software bill of materials, lists the software components in a product. An AI BOM extends that list to the parts that only AI systems have, such as models, training datasets and prompts. The software libraries still appear, so an AI BOM usually includes what an SBOM would hold.
What goes in an AI BOM?
An AI BOM lists the models with their versions, the datasets used to train or tune them, the software libraries and the configuration. Newer formats also record prompts, AI agents and the tools they can call. Each entry notes where the part came from and the licence it carries.
Is an AI BOM the same as an AI inventory?
No. An AI BOM describes the parts of one AI system, written by the team that built it. An AI inventory lists all the AI in use across a company, including the agents, MCP servers and skills employees installed themselves. A company can hold both, and each answers a different audit question.