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.
Transparency is useful when it helps a person understand what they are dealing with and what they can do next. A label alone can be accurate while leaving the important questions unanswered.
Design disclosure around the decision
An illustrative AI support interface can explain that an answer was generated, identify the source material where appropriate and provide a route to a person when the answer is uncertain. Which disclosures are legally required depends on the system and context; the design question is how to make them useful.
Put information where it affects the person’s decision. A disclosure hidden in a long policy may not help someone deciding whether to rely on a particular answer. Equally, repeating a broad warning on every screen can make the useful distinction harder to notice.
Test comprehension rather than merely visibility. Can the person explain what the system did and what remains uncertain? Do they know how to challenge or correct the result? A ticked box records an interaction, not necessarily understanding.
Keep the language specific to the behaviour. “AI may make mistakes” is less informative than explaining that a recommendation has not been independently verified. Do not promise human review if the workflow cannot actually provide it.
| Decision | Evidence to request |
|---|---|
| Context | What the system did |
| Limitation | What remains uncertain |
| Next step | How to question or correct the result |
What to do with this
Good disclosure is part of the product experience. It should make a decision easier to understand rather than merely relocate responsibility into the smallest text on the page.
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.
