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.
There is no single AI compliance clock. Different obligations, roles and systems can have different dates. A single countdown on a roadmap can therefore be precise and still answer the wrong question.
Map the obligation to the system
Begin with the system and the organisation’s role. Then identify the potentially applicable obligation, its source and its application date. Keep exceptions and transition provisions attached to that assessment rather than assuming the headline date settles the matter.
An illustrative organisation may use a general-purpose model within a separate application. Questions about the model provider and questions about the application should not be collapsed into one row. The responsibilities need to be assessed in context.
Use a dated review record. If the official position changes, update the interpretation and explain which implementation decisions are affected. Preserve enough history for another reviewer to understand why the earlier plan looked different.
This is particularly important when communicating externally. A statement that the product is “ready for the AI Act” can be much broader than the evidence supports. Explain the assessed scope, the controls tested and any unresolved questions.
| Decision | Evidence to request |
|---|---|
| Role | Which organisation does what |
| Obligation | Relevant source and assessed scope |
| Date | Application and transition assessment |
What to do with this
A calendar becomes useful after the scope is understood. Before that, it is mostly a way to schedule confidence. Keep the reasoning with the date.
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.
