The AOS Foundation advances the mission of building open standards, deterministic governance, and constitutional frameworks for autonomous AI agents.
The next generation of AI is autonomous. Models don't just answer questions — they take actions, manage infrastructure, write code, and make decisions that affect the physical world.
The AOS Foundation was established to ensure this transition happens safely, transparently, and under verifiable governance.
We believe that AI governance cannot be enforced by language alone. A model can be instructed to "be safe," but instructions are probabilistic — they can be circumvented, ignored, or misinterpreted.
The Foundation develops and maintains deterministic verification standards — governance enforced by code, not by prompts.
The principles that guide every initiative, standard, and tool produced by the Foundation.
Safety must be enforced by code, not by language. Every critical agent action must pass through a deterministic verification gate before execution. No exceptions.
The user is sovereign. They retain the right to inspect, fork, modify, and terminate any agent. They own all data generated. This is technically enforced, not merely promised.
All agent reasoning must be logged to an immutable, auditable ledger. No unlogged actions, no side channels. Every decision is traceable from intent to execution.
All Foundation standards require that agent actions serve a defined humanitarian purpose — uplifting sovereignty, protecting dignity, increasing access to knowledge, or reducing suffering.
Active projects and standards maintained by the Foundation.
The open standard for verifiable AI safety. A deterministic verification layer between agent intent and execution.
aos-governance.com →The constitutional framework that defines the rights, restrictions, and verification requirements for all AOS-compliant agents.
aos-constitution.com →The public evidence repository. Tamper-evident records verifying the development timeline, decisions, and governance of AOS.
aos-evidence.com →The specification for deterministic verification gates. Defines the Intercept → Verify → Gate pipeline for agent action governance.
Five non-negotiable principles governing all AOS-compliant agents: Humanitarian Purpose, Verification Gate, User Sovereignty, Kill Switch, and Transparency.
A portable, substrate-agnostic format for packaging governance capabilities as installable agent skills. Designed for interoperability across AI platforms.
The protocol for anchoring governance decisions to tamper-evident audit trails, enabling public verification of the entire decision chain.
The age of autonomous AI requires new institutions. The AOS Foundation is building the governance infrastructure for a world where AI agents act on our behalf.
Your privacy, your choice
We use cookies to enhance your experience. Essential cookies are required for the site to function. You can customize your preferences below.