NTT Docomo Business Develops Prototype AI Agent Registry

NTT Docomo Business, the enterprise services division of Japanese telecommunications group NTT Docomo, has developed a prototype registry system designed to establish the identity and verify the attributes of autonomous AI agents participating in digital transactions and multi-agent service interactions.

The prototype, called the AI Agent Attribute Information Registry, is designed to address a trust problem in emerging AI-agent ecosystems: existing identity and access management systems were not built for agents that are temporary, autonomous, distributed across platforms, and capable of acting independently at machine speed.

The registry uses verifiable credentials and digital certificates aligned with World Wide Web Consortium (W3C) Verifiable Credentials standards to authenticate AI agents and validate a structured set of attributes. These attributes include ownership (who operates the agent), operating authority (what organizational mandate it carries), execution permissions (what actions it is allowed to take), operational environment (what infrastructure it runs on), and data usage policies (what data it may access or process). The system is described as designed to detect impersonation attempts, unauthorized modifications, and fraudulent agents.

The technical centerpiece of the prototype is the AgentCard, a digital identity profile linked to verifiable credentials that agents present when interacting with other agents or services. AgentCards are described as persistent, tamper-evident identifiers. The registry is designed to interoperate with the Agent2Agent (A2A) protocol, a framework enabling multi-agent coordination where autonomous agents independently communicate, exchange data, and execute tasks. NTT Docomo Business plans future enhancements including integration with digital identity wallets and distributed key-management systems, incorporating Decentralized Identifiers and cryptographic signatures.

The AgentCard approach to agent attribution parallels World ID’s Agent Kit for iris-verified agent identity, which also uses W3C verifiable credentials to bind autonomous agents to accountable human principals. The registry’s planned DID integration aligns with W3C’s updated DID v1.1 working draft published in February 2026. At the regulatory level, NIST opened a public comment period on AI agent identity and authorization standards in March 2026, identifying the same gap in existing IAM frameworks that the registry prototype addresses.

NTT Docomo Business’s prototype work runs alongside related NTT Group AI initiatives, including NTT Docomo’s commercial deployment of an agentic AI system for network maintenance announced in February 2026 and the SyncMe personal AI agent pilot shown at MWC Barcelona 2026.

Sources: IT Business Today, Telecompaper, NTT Docomo

–

By the ID Tech Editorial Team