The institutional surface for governed AI · sealed receipts only.
No receipt, no governed decision

The institutional surface for governed AI.

The major laboratories sell mythos. We publish receipts.

GovernedAI defines, tests, and registers the operating doctrine for governed AI decisions: signed, replayable, hash-anchored artifacts that record what an AI system did, why it was authorized, what evidence was used, what was suppressed, and how to reproduce the decision bit-for-bit.

No receipt, no governed decision.
No verifier, no receipt.
No conformance, no category.
Doctrines without software counterparts are aspirations.

Five Pillars

Governed AI rests on five peer substrates. Each is independently citable.

ARS-OMEGA
Cognitive & Kernel Governance Substrate
Hardware-rooted enforcement of cognitive-layer decisions via signed Capability Tokens.
Enforces
PEL
Governed Process Substrate for AI cognition
Software-enforced operational discipline running closed improvement loops on attested evidence.
Improves
RDL
Decision Evidence Substrate
Open protocol for signed, replayable, hash-anchored decision artifacts.
Records
GDTK
Conformance Substrate
Public 100-case test kit that grades AI systems on the G0–G7 ladder.
Tests
Council
Adversarial Governance Substrate
Multi-AI governance with Sovereign / Builder / Auditor / Researcher / Adversary roles.
Audits

What is a governed AI decision?

A consequential AI output that emits a signed, replayable, policy-bounded artifact recording its authority context, inputs, configuration, evidence, suppression history, review path, and verifier state. Read the definition.

Why now

The market does not need another governance framework. The market does not need another observability dashboard. The market needs software-enforced discipline that turns existing frameworks and observability into closed, audit-grade improvement loops on hardware-anchored evidence.

Three forcing functions converge. First, full EU AI Act applicability begins in August 2026 (Regulation 2024/1689, Article 113). Second, large-enterprise procurement is expected to require ISO 42001 alignment from AI vendors as part of governance maturity expectations during 2026–2027. Third, the insurance and audit markets are starting to reflect AI governance maturity in their pricing and review terms. The vendors who can produce receipts will pass. Those who cannot will discover that mythos does not satisfy a regulator with a fine schedule.

Source receipts for the second and third forcing functions are being assembled in /receipts/. Until those receipts publish, the language above is qualitative — we name the direction, not specific numbers.

Recent receipts

The receipts registry is being seeded as Phase 1 launch artifacts pass adversarial review. Forthcoming entries: GDTK-100 specification, RDL Protocol v0.1, the KILL-DEAL audit narrative, the AIGP × C2PA × Microsoft AGT reconciliation paper.

Visit the receipts registry →

The Council ledger

This publication is itself governed. Read the governance model, including the Council Document-Parity Rule that will be ratified into AI Governance v2.3 at the 2026-06-30 quarterly review.

Doctrines without software counterparts are aspirations