AI auditability risk check
Could your team reconstruct one AI-assisted decision six months from now?
Answer a closed-ended self-assessment to see whether an AI workflow has an evidence gap worth stress-testing.
Keep this public self-assessment free of sensitive data.
This simulator is for educational purposes only. It is not legal advice, audit assurance, compliance certification, or a substitute for professional review. This page collects no free-form workflow narrative, customer data, regulated data, privileged material, or internal system details.
Self-assessment
Check one AI workflow against the eight evidence elements.
The useful answer is the defensible one. Select what you could show today, not what you expect the system should be able to show.
The workflow
Choose the environment and workflow you want to pressure-test.
Do you currently have an AI or GenAI workflow in production or serious pilot?
Do you have an internal audit, risk, compliance, legal, or AI governance function?
Which workflow are you most concerned about?
Auditability
For a specific past AI output, could you produce each of these today, on demand, with evidence?
Identify which model version produced a specific AI output
Recover the exact prompt or system instruction used
Show the input data submitted to the system
Show which sources, documents, or records were retrieved
Show the AI output as it was originally generated
Show who reviewed or approved the output
Show the approval and override record
Show where the evidence is retained
Readiness
Name the external pressure without submitting sensitive details.
Has an audit committee, regulator, customer, investor, or executive asked how your company governs AI?
Is there a real event or deadline driving this?
Are you interested in a fixed-scope 10-business-day sprint at $9,500?
Before you scale
Before you scale AI, test whether you can defend it.
The auditability gap is already here. The only question is whether you find it first - or a regulator, auditor, customer, or litigator does.