Responsible AI Platform
Back to the monitor

AI Act Monitor

Draft provides examples for high-risk classification

Detailed Commission guidance with examples for each route is now available, but its legal status remains draft.

Draft guidanceImportant development

Source facts

What officially changed?

On 19 May 2026, the Commission published three draft guidelines on the classification of high-risk AI. The consultation has closed, but the guidelines have not yet been finally adopted.

Legal effect

The draft explains the Annex I and Annex III routes and the exceptions under Article 6(3). Dates in the draft must be read together with the later binding postponement in Regulation (EU) 2026/1744.

What the status means

This is draft guidance. The final text may still change, so use it as a preparation signal rather than a final position.

Analysis by Responsible AI Platform

Why is this material?

Detailed Commission guidance with examples for each route is now available, but its legal status remains draft.

What does this mean in practice?

The examples are useful for a provisional review of an inventory. A classification decision must still be substantiated under Article 6 and the annexes, noting that the final guidelines may differ.

What may need to be reassessed?

  • Provisionally test existing classifications against the Annex I and Annex III examples.
  • Record assumptions and any reliance on Article 6(3) explicitly.
  • Use 2 December 2027 and 2 August 2028 as the applicable dates.
  • Schedule a reassessment when the final guidelines are published.

For whom and which provisions?

Perspectives

Using AI, Providing AI, Advising, Public authority

Topics

High-risk AI, FRIA

Legal evidence

Official sources

Last checked by Responsible AI Platform: 8 August 2026.

Only when something changes

Follow the topics in this change

Receive only material changes that match this selection. No more than once a week.

This analysis provides general information and is not legal advice for a specific situation.