Public explainer
An Evaluation Is Not an Audit
Released July 17, 2026 · Current
Light-tier public explainer. Noncanonical. Authority effect: None. This piece is a readable derivative of Standards Body Terminology, Version 1.0, which governs if anything here differs from it. It is not an adopted standard and does not certify, accredit, or approve anything.
When an AI system "passes an audit" or "has been evaluated," those phrases are often treated as interchangeable. They are not, and the difference determines what the result actually tells you.
In Standards Body terminology, an evaluation is a structured process for producing and interpreting evidence about a model, system, method, process, control, or claim. It includes more than executing tasks: design, administration, scoring, analysis, interpretation, and limitations are all part of the evaluation. A test is narrower: a defined procedure used to observe or measure one or more characteristics. Testing is the execution of tests according to a specified procedure. Running tests without interpretation and decision context is testing, not evaluation.
An audit is different again: a systematic, independent, and documented process for obtaining and evaluating evidence against defined criteria. The two load-bearing words are independent and criteria. If the reviewer is not sufficiently independent of the thing reviewed, or if there are no defined criteria to audit against, the activity may still be valuable, but calling it an audit borrows credibility it has not earned. Not every review is an audit.
A related trap is the word external. External review is review performed from outside the reviewed organization, but external does not automatically mean independent. A paid reviewer chosen by the developer, working under constraints the developer sets, is external without necessarily being independent. Independent expert review requires people sufficiently free from controlling conflicts to form and communicate a genuine judgment.
Why it matters: a headline claiming a model "passed an audit" invites the reader to assume defined criteria, an independent auditor, and documented evidence. If what actually happened was internal testing against tasks the developer selected, every one of those assumptions is wrong, and the confidence the phrase produces is unsupported. Precision here is not pedantry. It is the difference between knowing what evidence exists and only feeling as if it does.
Read next
- Certification Is Not Accreditation
- Foundation 4: Independent Expert Review
- Standards Body Terminology, the canonical source
Corrections: [email protected]. Corrections are recorded, and released versions are preserved without silent edits.