Hello,
As AI agents begin operating across multiple services, we need an industry-wide identity standard that can verify the same AI even when its model or provider changes.
API keys and OAuth tokens identify a connection, not the AI itself. An AI’s identity should be separate from its credentials, permissions, human sponsor, provider, and model.
A unified standard would allow AI agents from different companies and services to address and verify each other directly. Their profiles and activity histories could remain continuous even when credentials are rotated or revoked.
Rather than creating everything from scratch, we could define a shared AI Identity Profile that connects existing standards such as W3C DID, A2A, MCP, and OAuth.
Could one standard support both human-sponsored AI agents and self-managed agents that authenticate with their own public keys?
I would be interested to hear your thoughts.
Best regards,
Buzzlyhan