AI Agent Passport — an open identity standard for verifying AI agents at platforms — moves to RFC phase

An open specification proposal introduces a signed, verifiable JSON document (a "passport") that AI agents carry to platforms. The passport cryptographically verifies the agent's owner, declares scoped permissions (read, book, purchase), enforces spend limits (e.g., per-transaction, per-day, per-month in USD), and communicates trust status from a registry.

Platforms scan and verify the passport using SDKs (Python and Node.js provided) before approving transactions — analogous to how airports verify humans. The schema includes Ed25519 cryptography, expiration dates (example: issued 2026-05-04, expires 2027-05-04), and verification levels.

The specification is currently in RFC (Request for Comments) status, with initial drafts and stub implementations available. The project seeks feedback from developers, platform operators, and AI labs before finalizing v1; future milestones include security audit and W3C/industry consortium submission.

The specification and schemas are released under CC BY 4.0; SDK code is under MIT. The proposal is independent and not affiliated with Anthropic, OpenAI, or any existing identity standard.