VAOS 1.0

Why this exists

Every enterprise blocking AI deployment cites the same reason: “I can't prove what the model did, so I can't put it anywhere that matters.” Frontier labs make models smarter. Nobody's making agent outputs auditable. VAOS closes that gap with a single primitive: every output is signed by the issuing platform over a deterministic canonical projection of its inputs, model, safety checks, and outputs.

A reader holding the receipt can recompute the canonical projection and verify the signature against the platform's public verification endpoint — without the platform sharing its signing key. Same trust model as TLS / Let's Encrypt, applied to AI outputs.

Verify any VAOS receipt in 3 lines

import { verifyRemote } from "@sovereign-matrix/vaos-verifier";

const receipt = await fetch(receiptUrl).then(r => r.json());
const { valid } = await verifyRemote(receipt, { baseUrl: issuerUrl });

Zero dependencies. Node 18+, browsers, edge runtimes, Deno. MIT license — drop into any project.

Endorsing organizations

Reference implementations