OpenMatter and Hashgraph Online Propose AI Agent Compliance Standard

OpenMatter Network and Hashgraph Online, a standards group built on the Hedera network, published a draft specification on GitHub for enforcing AI agent policies and verifying compliance without revealing sensitive data. The organizations opened the document for public comment through October 31, 2026, with a possible extension.

The proposed standard, titled Zero-Knowledge Boundary Compliance for Autonomous Agents, defines requirements for controlling what agents can access, share and do when interacting with users, data and external systems. Systems following the specification could block actions that violate a policy and produce independently verifiable evidence that permitted actions complied with the rules.

OpenMatter, Hashgraph Online, Vera Anchor and other members of Hashgraph Online's AI Privacy & Security subcommittee developed the draft. After the comment period, the working group will review feedback, document how substantive comments were resolved and publish a revised version with updated compliance requirements and test vectors where appropriate.