What is an AI Audit Trail?
An audit trail for automated decisions is a record, for each decision, of what was decided, which rule decided it, the facts the decision rested on and which version of the rules was in force. It is what lets anyone answer, later, why a decision came out the way it did.
What NOMOS records
Every answer names the rule that decided it, the facts it was given and the sealed version, and the records are hash-chained so that a removed or altered entry breaks the chain. Answers from the Exchange are also signed with the platform's published key, and can be checked offline.
What the law asks for
Under the EU AI Act, high-risk AI systems must technically allow for the automatic recording of events over their lifetime (Article 12(1)), and deployers must keep those logs for at least six months unless other law provides otherwise (Article 26(6)). A record of decisions taken under the rules is evidence alongside that logging; it does not replace it.
Example
A customer disputes a declined loan eighteen months later.
The model has been retrained twice. Nobody can say which rules applied on the day.
The record shows the rule that declined it, the facts it was given and the sealed version in force that day, which still verifies.
Frequently asked questions
Related concepts
See it in practice
Upload a policy, confirm its rules, seal them: any system can then check an action against them.