AI Audit Exchange
Go back

Guide · By the Ingenious AI team · Published 14 November 2026 · Reviewed 14 November 2026

Stale Audit Evidence: The 365-Day Freshness Rule Explained

What is the freshness rule?

The freshness rule is the stated product rule applied by AI Audit Exchange's room validation: an artifact is flagged stale when its recorded date falls more than the freshness window — 365 days — before the validation run. Validation results are stored per room and shown with the rule itself stated on the surface, so anyone reading a validation can see exactly which standard produced its flags.

Why evidence has a freshness problem at all

An audit examines a period, and evidence testifies about practice during a period. A control record from three cycles ago may be perfectly authentic — sealed, badged, hash-verified — and nearly useless as testimony about how the control operates now. Organisations change tools, staff and configurations constantly; evidence ages at the speed of those changes. The freshness rule exists to make ageing visible before an auditor does. Without it, a room can look complete by count while its centre of gravity sits in the distant past. With it, the monthly question becomes precise: which artifacts are old, and does current evidence exist to stand beside them?

The rule, exactly as the platform states it

Validation reads each artifact's recorded date and compares it with the run's date: more than 365 days between them, and the artifact is flagged stale. The rule is stated on the room's surface wherever validation results appear — the window, the comparison, the outcome — because a flag whose standard is hidden cannot be argued with, cleared intelligently or explained to an auditor. Note what the rule uses: the artifact's recorded date, the date the evidence itself carries. That is the date that matters for testimony. The import date is recorded separately in the manifest and tells the filing story; the freshness rule interrogates the evidence's own age.

What a stale flag is not

Three misreadings to avoid. A stale flag is not an integrity problem — the hash still verifies; the file is unchanged, merely old. It is not a compliance verdict — the platform makes no legal determinations, and some regimes and questions legitimately reach back years. And it is not an instruction to delete — the old artifact stays, sealed, as part of the record; the correct response is almost always to add current evidence, not to remove old evidence. Teams that treat flags as deletions end up with rooms that are fresh and thin. Teams that treat them as prompts end up with rooms that are deep and current.

Clearing flags the right way

Work the flag list as a filing list. For each stale artifact, ask what its current-period equivalent is: the same product's export for this period, the same control's record from this cycle. Import that equivalent through the proper path — suite exports unchanged through import, organisational material as direct uploads — and the next validation run shows the room's centre of gravity move forward. Some flags will remain, correctly: a founding policy document, a long-lived architectural record. Those you acknowledge individually, in the room's question thread or your review notes, with the reason they are old by nature. An auditor who finds three explained old artifacts and forty current ones reads a maintained room.

Freshness across regimes

The rule is regime-neutral; its weight is not. In a CPPA room, where the audit is a cybersecurity audit, staleness goes to the heart of the examination: security practice is current or it is historical, and last cycle's records testify weakly about this cycle's controls. Weight CPPA flags heavily and clear nearly all of them. In an SB 1119 room examining behaviour across a long period, older artifacts may be the substance of the examination itself. Same flags, different reading — the rule surfaces the facts, and the regime supplies the interpretation. The contrast is drawn out in SB 1119 vs CPPA.

Making validation a rhythm, not an event

Run validation monthly during the period and always once before the auditor starts. Monthly runs turn freshness into a trend: the stale count should fall as the period's filings land, and a rising stale count is an early warning that filing has stalled while time has not. Store each run's results with the room — validation is part of the room's working record, visible alongside its questions and access log. By the time the auditor runs their own eye over the room, freshness should be a topic you have already managed for months, with the runs to prove it.

Five actions to take

  1. Run validation monthly and once before the auditor's start date — treat the stale list as a filing worklist.
  2. For each stale artifact, import its current-period equivalent through the proper path; never delete old evidence to clear a flag.
  3. Acknowledge legitimately historical artifacts individually, with the reason recorded in the room.
  4. Weight CPPA staleness heavily — cybersecurity evidence must speak for current practice.
  5. Track the stale count as a trend across runs; a rising count means filing has stalled.

Worked example

A fictional travel-platform company, Blue Agave Travel Tech, runs validation on its CPPA room of 66 artifacts in September. Nine flags come back: seven artifacts recorded more than 365 days earlier. Working the list, the team finds five have current-period equivalents that were produced but never imported — a filing gap, now closed with 5 imports. Two are old configuration records whose systems were replaced in March; the March replacements are imported as 2 new artifacts and the old pair acknowledged in the room with a note.

The October run flags 2 artifacts, both acknowledged historicals. When the auditor asks about freshness — auditors always ask about freshness in a cybersecurity examination — the room shows two runs, a worked list and a current evidence base. The conversation moves to substance in the first meeting, which is where a prepared auditee wants it.

Frequently asked questions

What date does the freshness rule use?

The artifact's recorded date — the date the evidence itself carries — compared against the validation run's date. More than 365 days between them flags the artifact as stale.

Does a stale flag mean the artifact failed its hash check?

No. Freshness and integrity are separate properties. A stale artifact's hash still verifies; it is authentic but old. Validation reports freshness; hash verification reports sameness.

Should I delete stale artifacts?

Almost never. The record's depth is valuable, and deletion is governed by retention and hold, not by flags. Add current evidence alongside; delete nothing to make a report prettier.

How often should validation run?

Monthly during the audit period, and once more immediately before the auditor begins. The trend across runs matters as much as any single result.

Who sees validation results?

The room's owner and its granted auditor. Validation is stored per room as part of its working record, with the freshness rule stated on the surface beside the results.

Keep reading

All AI Audit Exchange guides · Auditor directory · About AI Audit Exchange · Security · Pricing · AI Audit Exchange home · Previous: What Is an Evidence Completeness Checklist? · Next: Auditor Questions and Answers Inside the Evidence Room

How Auditees and Auditors Share Evidence Securely · Evidence Room vs Shared Drive: An Honest Comparison · How to Organise AI Audit Evidence by Period