Proving AI decisions to your board: a 1-page evidence checklist
The minimum evidence an audit, risk or compliance committee should expect for every automated decision - signed, chained, time-anchored and verifiable offline.
About the author+
Jamil Luketic
Executive Director at Decision Keep
Former Data & Tech Leader at Oracle, Mastercard, Coles, Optus, and Reece.
Connect on LinkedInMost boards do not need a 40-page AI policy. They need one page of evidence that answers a single question: if someone challenged a decision our AI made, could we prove what happened - and that nobody changed the record? Gartner, PwC, Deloitte and EY all now publish board-level AI oversight guidance to the same effect.
This checklist is that one page. Share it with your audit, risk and compliance committee. If every box is ticked, your organisation can answer "yes" with confidence.
The 1-page evidence checklist
For every automated decision your organisation makes, you should be able to show:
- Signed with your key. Each decision is sealed with your organisation's own Ed25519 key, not the vendor's. The signature proves origin and integrity.
- Hash-chained. Every record links to the one before it. Tamper with any entry and the break is detectable by re-checking the chain - no database required.
- Time-anchored. An independent RFC 3161 timestamp proves when the decision was signed.
- Logic captured. Model version, relevant inputs (references, not raw PII), confidence/risk scores and the routing outcome are recorded.
- Verifiable offline. An auditor can verify a receipt against your published key with no account and no trust in the platform.
- Sovereign. The evidence lives in an open format on infrastructure you control.
- Erasable, provably. When retention expires, content is cryptographically erased while a signed proof keeps the chain anchor.
These six properties are explained in depth in The Forensic Witness: why automated decisions need an independent record.
Why "trust the dashboard" is not enough
A live dashboard shows how a system is behaving now. It does not prove what a specific decision was at a specific time, or that the log behind it hasn't been edited. When a regulator asks, the dashboard is a snapshot; the audit trail is the record.
A record you can only inspect by logging into the vendor's console is not auditable. Auditable means an auditor can prove your claims against you.
Using this with Decision Keep
Decision Keep produces exactly this evidence for every automated decision: signed, chained, time-anchored, and verifiable offline against your published key. Each record is sovereign to your infrastructure, so the board's evidence never leaves your control.
- Verify a sample receipt with no account.
- Read the receipt specification.
- Book a demo to map this checklist to your stack.
FAQ
Questions auditors, risk and legal actually ask
What AI evidence will the board ask for?+
What is the minimum evidence a board needs for AI decisions?+
Why should the evidence be verifiable offline?+
Does this satisfy the EU AI Act and APP 1.7–1.9?+
Sources
References & further reading
Independent analysis and standards cited in this article.
- AI Regulations to Drive Responsible AI Initiatives
Gartner · 2024
- Trustworthy and Ethical AI Thought Leadership
Deloitte · 2024
- Trust AI: operationalizing trust at scale
PwC · 2024
- Assuring trust in an AI-driven world
EY · 2024
Prove every AI decision
Decision Keep gives your organisation a tamper-evident, verifiable record of every automated decision. Book a demo to see it on your stack.
Keep reading
Authorization decision logging: how fast AI risk scoring stops an authorization decline spike
An authorization decline spike is the worst kind of payment event: revenue hemorrhages, customers complain, and every denied transaction is now a potential d…
What tool can automatically identify at-risk accounts before they cancel?
A churn model can flag an account likely to cancel in milliseconds. But the flag itself the automated decision to treat this customer differently is the thin…
AI-native vs AI-enhanced risk decisioning: where the evidence gap widens
Most organisations can tell you whether an AI system is "enhanced" or "native." Far fewer can prove what each decision was , and that nobody changed the reco…
Documentation