PCAOB · AU-C 500 · AI-Tool Evidence

AI Evidence Auditors Can Actually Review

Amended AU-C 500 asks firms to document six things about an AI or technology-assisted tool used in an engagement: what the tool is, how it was used, what it was given, what it produced, its limitations, and who signed off. A signed OpenChainGraph compute receipt already carries every one of those, in a form a reviewer can independently replay — not take on faith. This guide walks through the mapping, then a real replay a reviewer can run in their own browser.

Six Documentation Elements Independently Replayable Zero PII Evidence, Not Opinion
๐Ÿ”’ All inputs are processed locally in your browser. No data is transmitted. Do not enter real personal data — use synthetic or anonymised inputs only.
Evidence, not opinion: the tools linked here are designed to satisfy the documentation elements of AU-C 500 (AI/technology-assisted evidence, as amended), AS 1105 / AS 2301, and QC 1000 / AS 1215 — with citations, versioned. They carry no PCAOB or AICPA endorsement and assert no compliance sufficiency. Auditors form opinions; a receipt is something an auditor can check.
The Mapping

Six elements, one receipt

Every field below is populated from a real signed OCG compute receipt — nothing here is narrative the auditor has to trust; each element is a value the reviewer can check against the receipt directly.

01 · Tool Identity

What the tool is

The receipt's kernel_id and engine digest identify the exact deterministic compute kernel that ran — not a product name, a specific pinned version.

02 · Extent of Use

How it was used

The record composed by T561 states the analysis performed, in the auditor's own description, bound to the receipt it accompanies.

03 · Input Reliability

What it was given

Inputs are represented as digests, with optional attestations where the source can be observed. Internal-tool and external-tool inputs are labeled differently — an external analysis is asserted, not observed.

04 · Output

What it produced

The output payload's digest is bound into the receipt's execution_hash — a reviewer verifies the stated output is the actual output, not a description of it.

05 · Limitations

What it doesn't claim

Expressed as a determinism class plus declared conventions — not as reassurance. A receipt says what was computed and how reproducibly, nothing more.

06 · Sign-Off

Who reviewed it

A countersignature slot binds a named reviewer's attestation to the receipt, composed by art-380 (build_ai_workpaper_record).

AU-C 500, as amended AS 1105 / AS 2301 QC 1000 / AS 1215
Replay Walkthrough

Verify a receipt yourself, in-browser

A reviewer does not have to trust the claimed execution_hash on a workpaper record — the Kernel VM Widget re-runs the same kernel in a sandboxed, deterministic VM and computes its own hash to compare against.

1
Produce or receive the record
Run T561 (technology-assisted-analysis record) or T562 (engagement-documentation gap check), or take a workpaper record already composed via art-380. Each carries a kernel_id, policy_parameters, an output payload, and a claimed execution_hash.
2
Open the Kernel VM Widget
Go to Kernel VM Widget. It runs a ChainGraph kernel's compute(policy_parameters) inside a sandboxed, deterministic in-browser VM, independent of the page that produced the original receipt.
3
Paste the same inputs
Enter the record's kernel_id and policy_parameters exactly as they appear on the receipt. The widget re-executes the kernel from those inputs alone — it does not read the claimed output.
4
Compare execution_hash
The widget shows its own independently-computed execution_hash alongside a native (non-sandboxed) re-run in the same tab. If both match the receipt's claimed hash, the record's output is exactly what the kernel produces from those inputs — not a description the auditor has to take on faith.
Related

Explore adjacent tooling