The Evidence Standard Is
Not a Compliance Exercise

Regulators, auditors, and counterparties are converging on the same question. The organizations that can answer it are in a structurally different position than those that cannot.

The Composition Gap

Two or more individually authorized AI agent actions combine to produce an outcome the governing organization has not authorized. No governance mechanism evaluates the combination as a unit before any consequential action executes.


This is not a failure of individual action controls. It is a structural condition that exists in every sequential AI system that evaluates individual action permissions without evaluating the combination as a unit. Existing runtime monitoring, IAM, and guardrail systems address individual actions. None evaluate the sequence.


The Composition Gap predates AI. AI accelerates it. The regulatory disclosure requirements, the securities litigation, and the national security guidance on agentic AI are all responses to the same structural condition at different layers.

Action A Individually permitted
+
Action B Individually permitted
+
Action C Individually permitted
=
Combined Outcome Organization never authorized

No existing control evaluated the combination before Action A executed.

Four jurisdictions. One evidence standard.

European Union

EU AI Act · Articles 12 + 13

Article 12 requires

Logs enabling reconstruction of the circumstances leading to any output. The record must enable a regulator to understand what happened without the controller's cooperation.

Article 13 requires

Transparency about the data and knowledge the system was acting on at the time of the decision. The knowledge state must be recorded, not reconstructed from memory.

Provisional agreement reached May 7, 2026 delays Annex III enforcement to December 2, 2027 and Annex I to August 2, 2028. Formal adoption pending. The evidence standard is unchanged regardless of timeline.
United States

RAISE Act · Effective January 1, 2027

What it covers

Large frontier model developers with $500M+ revenue and models trained at 10²⁶+ FLOPs. Safety protocols, incident reporting within 72 hours, and pre-deployment safety testing.

The gap it leaves

Deployer-side governance — what organizations must demonstrate about how frontier models were authorized to act in their specific workflows — is outside the RAISE Act scope. That gap is not addressed by any existing regulation.

United States · GSE

Fannie Mae LL-2026-04 · Effective August 6, 2026

Purpose disclosure

What was the AI deployed to do? Requires a record of the intended AI function at the time of the decision — not a policy statement about AI use in general.

Manner of use disclosure

How did the AI operate in this specific instance? Requires a record of actual execution behavior — not a description of model capabilities.

SOC 2 does not satisfy either disclosure requirement. Both require pre-execution records that existed before the AI acted in the specific instance.
Canada

OSFI E-23 · Effective May 1, 2027

What it requires

All federally regulated financial institutions must maintain model inventory, document model lifecycle governance, and demonstrate board-level oversight of AI systems. AI and ML systems are explicitly included.

The evidence question

Was lifecycle governance applied before the model acted — or assembled after the fact? The pre-execution authorization record is the evidence that governance was operational, not retrospective.

National security agencies have named
the structural problem.

CISA, NSA, and allied cybersecurity agencies from Australia, Canada, New Zealand, and the United Kingdom issued coordinated guidance on agentic AI systems in May 2026. The guidance identifies the same structural condition Kuriom addresses.

Accountability Risk

"Agentic systems obscure what caused a particular action, making accountability hard to trace." — CISA/NSA Five Eyes Guidance, May 2026

The KGIF immutable ledger produces a pre-execution record that enables exact reconstruction of what authorized the action, before the action occurred.

Structural Risk

"Multiple interconnected components that plan, reason, and act across sequential steps introduces new systemic risks including cascading failures." — CISA/NSA Five Eyes Guidance, May 2026

The Sequence Authorization Unit evaluates the complete sequence as an indivisible unit. Cascading failures across sequential steps are prevented before the first step executes.

The organizations that can answer the regulatory evidence question with a pre-execution governance record — not a vendor attestation, not a methodology document, but an independently verifiable record that existed before the AI acted — are in a fundamentally different liability position than those that cannot.