Updated 17 October 2026. This article has been revised to separate the policy timeline from the product argument. See the current AI Act timeline for the category-specific dates and primary reference.
A readiness percentage sounds precise. Without the sample, question and method, it may tell you very little about your organisation. Earlier numerical claims in this article were not adequately sourced and have been removed.
Read the survey before repeating the number
Ask who was surveyed and what they were asked. A respondent’s confidence in passing an audit is different from an independently assessed control result. An inventory question is different from a question about overall governance maturity.
Check the population and collection period. A survey of a particular industry or region should not quietly become a claim about all enterprises. A result from last year should not be described as a measurement of the current market without qualification.
Even a sound survey cannot establish the state of your own system. Use it to identify questions worth asking, then collect evidence locally. Can the team find the current inventory? Can it show a tested intervention? Can it reconstruct a sample action?
An illustrative internal review might mark each claim as demonstrated, failed or not yet tested. Keep the underlying evidence attached. Converting uncertainty into a neat percentage can be useful for reporting, but it should not erase the distinction between those states.
| Decision | Evidence to request |
|---|---|
| Survey claim | Question, sample and date |
| Local claim | System-specific evidence |
| Decision | Known result and remaining uncertainty |
What to do with this
A statistic can start a conversation. It should not finish the assessment. The useful question is what the team can demonstrate about the system it is actually responsible for.
Record the decision, the responsible owner and the evidence still missing. The decision-record example shows a lightweight way to do that. Where a claim concerns system behaviour, connect it to evaluation evidence. These are proposed product practices, not a finding that a system complies with the law.
Primary timeline reference: European Commission, AI Act. The operational recommendations above are my analysis.
