Governance Principles
Neutrality
No single company, platform, or government controls AIR. The standard belongs to everyone.
Transparency
All decisions, algorithms, and processes are documented publicly. Nothing happens behind closed doors.
Accountability
Published roadmap with honest status updates. We say what we've done, not what we hope to do.
Decision-Making
Specification Changes
Changes to the AIR specification follow a public process:
- Proposals are submitted as GitHub Issues or Discussions
- Community review period (minimum 30 days for significant changes)
- Technical merit and alignment with AIR's mission are the deciding factors
- All accepted changes are documented with rationale
Trust Score Methodology
The trust scoring algorithm is fully published. Any changes to scoring weights, components, or grade boundaries go through the same public review process as specification changes. No score adjustments happen without community visibility.
Operational Decisions
Day-to-day registry operations (infrastructure, security, uptime) are managed by the foundation team. Significant operational changes (new verification tiers, API breaking changes, data policies) follow the public review process.
Anti-Capture Provisions
AIR is designed to resist capture by any single interest:
- No corporate board seats — Board members serve as individuals, not as representatives of their employers
- Open source everything — Specification, scoring methodology, and registry code are all published under Apache 2.0
- No vendor lock-in — Built on W3C open standards (DIDs, VCs). Agent identities are portable by design
- Diversified funding — No single funder may contribute more than 25% of annual revenue (goal)
- Fork rights — If AIR ever fails its mission, the community can fork the specification and data
Verification Tiers
AIR uses a tiered verification system. Higher tiers require more evidence and produce higher trust scores:
| Tier | Requirements | Trust Impact |
|---|---|---|
| Self-Verified | Agent registered with a creator DID. No external verification. | Base score only. Limited trust ceiling. |
| Basic | Creator identity confirmed via domain verification (did:web) or email. | Score boost for provenance component. |
| Standard | Independent verifier attests to agent capabilities and behavior. Code repository reviewed. | Significant score improvement across multiple components. |
| Enhanced | Third-party security audit completed. Multiple independent attestations. Continuous behavioral monitoring. | Highest possible scores. Eligible for AAA grade. |
Future Governance Milestones
- Technical Steering Committee — Elected body to oversee specification evolution (planned for Q4 2026)
- Verifier Certification Program — Standards and process for certifying independent verifiers
- Community Advisory Council — Representatives from different stakeholder groups (developers, enterprises, researchers, civil society)
- Legal Structure — 501(c)(3) nonprofit filing (in progress)
Get Involved
Governance works best with diverse input. You can participate by:
- Joining GitHub Discussions — propose changes, ask questions, give feedback
- Reviewing the specification and trust methodology
- Emailing foundation@agentidentityregistry.org