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.
When a policy outcome is uncertain, teams can plan for more than one outcome without pretending to know which will happen. Scenario planning is useful precisely because prediction is unreliable.
Plan around decisions that may change
Start with the decision affected by the uncertainty. Is it the release date, the evidence required, a contractual commitment or the scope of a feature? A scenario that cannot change a decision is usually just a second slide deck.
For each plausible outcome, record the work that remains useful and the work that depends on that outcome. An accurate inventory and a tested permission boundary may be useful in every scenario. A specific filing process may depend on the final rule.
Use an official source and a review date for each assumption. When the status changes, update the plan. Do not keep a superseded proposal alive in a spreadsheet because the team spent a long time formatting it.
An illustrative project might prepare evidence while holding a disputed deployment decision for qualified review. That preserves progress without silently deciding the unresolved legal question.
| Decision | Evidence to request |
|---|---|
| Common work | Useful in every assessed scenario |
| Conditional work | Depends on a named outcome |
| Trigger | Official change that prompts a decision |
What to do with this
Keep the scenarios small enough to use. The objective is to avoid a surprise dependency, not to create a parallel universe for every possible press release.
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.
