Sui Announces Verifiable Agent Arbiter Launch for AI Oversight

Sui announced the launch of Verifiable Agent Arbiter, a system designed to help businesses check whether artificial intelligence agents act within the authority assigned to them. Developed with Google Cloud, the system is intended to provide an evidence layer for reviewing agent behavior in business operations.

The design gives its components separate roles: Walrus stores cryptographic proofs, while Sui coordinates verification. Detailed activity records are kept in customer-controlled Google Cloud Storage, so businesses can review an agent’s permissions, actions and results while sensitive operational records remain in private storage.

Sui presented the arbiter as a governance tool for AI systems operating in decentralized environments. Its purpose is to give businesses evidence they can use to check an agent’s activity against its assigned permissions. TokenPost previously covered the project’s planned verification design.