Policy Paper Calls for Shared AI Bill of Materials Standards
Summary
An Institute for Security and Technology policy paper says AI bills of materials could improve supply-chain transparency and reduce cyber risk, but adoption requires shared foundations plus demand through mandates, regulations, or contracts.
Key points
- The paper proposes AIBOMs as inventories of models, datasets, and other inputs used in AI development and operation.
- It argues that shared foundations are needed to avoid incompatible approaches and support interoperable tools and usable data.
- The authors recommend addressing both data supply and demand to resolve the adoption “chicken-and-egg” problem.
- Possible demand-side drivers include industry mandates, government regulations, and contracting conditions.
Article Details
- Event Type
- Publication of a policy paper proposing an approach for AI software bills of materials (AIBOMs).
- Impact
- The paper argues that AIBOMs could reduce cyber risk and improve transparency by tracking models, datasets, and other AI development inputs. It says broad adoption requires foundational work and both supply-side data collection and demand-side requirements, potentially through industry mandates, government regulations, or contracting conditions.