An AI-BOM, or AI bill of materials, is an inventory record for an AI system. It lists key artifacts such as models, datasets, weights, versions, provenance, and related trust data. Like an SBOM for software, it helps teams know what is inside an AI system, where each part came from, and which version is actually deployed.
This matters because AI supply chains are often assembled from multiple sources and can change over time. Security teams use AI-BOM data to detect undocumented models, verify chain of custody, and support incident response, audits, and change control. In real attacks, missing lineage can hide tampered models, swapped weights, poisoned training data, or shadow deployments. In defense, signatures, provenance records, and current inventories help confirm integrity. An AI-BOM improves visibility, but it does not prove the model is safe or trustworthy by itself.