Skip to content

RFC-010: Autonomous Intelligence Governance Protocol (AIGP) — 14. Summary

AIGP SpecificationRFC-010: Autonomous Intelligence Governance Protocol (AIGP) › 14. Summary

← 13. Current Implementations · Section index · 15. Agentic Governance Extension →

14. Summary

AIGP is a consent-based, HMAC-authenticated governance protocol that enables centralized oversight of AI usage across multiple applications and cloud providers. Applications register, check before invoking, and record after invoking. The governance authority (governance-server) operates in REPORT mode (observe and log) or ENFORCE mode (actively block violations and configure provider-native controls). ENFORCE maps AIGP policies to cloud-specific services Bedrock Guardrails on AWS, Content Safety on Azure, Responsible AI on GCP through a provider adapter interface. The protocol is cloud-agnostic; enforcement is cloud-specific.


← 13. Current Implementations · Section index · 15. Agentic Governance Extension →