A passport-based identity system for AI agents on the public internet.
Open protocol, edge-verified, sub-50 ms, vendor-neutral, chain-agnostic — built so legitimate agents pass bot defense, the rest stay blocked, and AI-generated content is cryptographically attributable.
Apache 2.0 · open spec · interop with C2PA, MCP, A2A, Kite, SPIFFE, OWASP ANS
Infrastructure for the agent layer of the internet
Bot operators get their agents accepted. Sites publish what bots may do. Bot-defense vendors get a verified-bot signal. Publishers sign their content. Compliance teams get machine-readable AI attribution. Everyone else gets a cleaner internet. Same protocol from every side — pick yours.
Stand up a tier-1 issuer in 5 minutes. Your bots present a passport; sites that recognize the protocol let them through. Anthropic, OpenAI, Cursor, Replit, Perplexity, Microsoft Copilot — anyone shipping agentic AI.
For agent builders →Publish a policy at a well-known URL. Honest shopping assistants pass; scrapers and engagement farms don't. Marketplaces, ticket platforms, review sites, dating apps, social platforms.
For site operators →Your detection engine gains "verified bot with declared intent" as a distinct class. Cloudflare, Akamai, HUMAN, DataDome, Arkose, Kasada — partner channel for the verifier.
For bot-defense vendors →Sign your article, image, video at publication. AI clones can't fake the signature; readers, search engines, and courts can verify. NYT, BBC, Reuters, Substack writers, photographers, news agencies.
For publishers →EU AI Act Article 50, China AIGC, US state labeling laws. AgentPKI Provenance turns disclosure obligations into cryptographic proof. AI labs, regulated enterprises, platforms operating in regulated jurisdictions.
For compliance teams →You don't deploy anything. The internet you use just becomes cleaner: bot intent gets declared, AI content gets attributed, scams and impersonations get cryptographically visible.
For everyone →AgentPKI is open protocol with a hosted reference implementation. Issuers, verifiers, policy holders, content publishers, regulators, agent builders, vendors — if you have a use case, I want to talk.
Or email [email protected] directly.
