Skip to main content
Praxikon

Comparisons

These pages put two provisions next to each other that run together in practice. What is in the table comes from the same versioned objects as the obligation pages, so you read the real fields per side rather than a summary of them.

  1. Article 16 versus Article 26: what the provider owes and what the deployer owes

    The duties of the provider of a high-risk AI system next to those of the deployer, with the locator, the duty holder, the evidence and the first actions per side, from the knowledge graph.

  2. Article 26 versus Article 27: using and assessing

    What every deployer owes, and what a subset of them owes on top of that. Per side the locator, the duty holder, the evidence and the first actions, from the knowledge graph.

  3. Article 4 versus Article 14: literacy and human oversight

    The duty to take measures around people next to the design requirement around systems. Per side the locator, the duty holder, the evidence and the first actions, from the knowledge graph.

  4. GPAI versus high-risk: the model or the use

    Two regimes that exist alongside each other and touch different parties. Per side the locator, the duty holder and the evidence, from the knowledge graph.

  5. Article 5 versus Annex III: prohibited or high-risk

    The difference between a practice that is not allowed and a use that is allowed under conditions. Per side the locator, the duty holder and the evidence, from the knowledge graph.

  6. Article 50 versus Article 13: two kinds of transparency

    Transparency towards the person who encounters AI, next to transparency towards the organisation that deploys the system. Per side the locator and the evidence, from the knowledge graph.

  7. Article 11 versus Article 13: technical documentation and instructions for use

    Two documents from the same provider, with a different reader and a different purpose. Per side the locator, the piece of evidence and the first actions, from the knowledge graph.

  8. Article 27 versus Article 9: FRIA and risk management system

    Two assessments of risk, held by different parties and with a different subject. Per side the locator, the duty holder and the evidence, from the knowledge graph.

  9. Article 14 versus Article 26: enabling oversight and assigning oversight

    Human oversight is a design requirement for the provider and an organisational duty for the deployer. Per side the locator and the evidence, from the knowledge graph.

  10. Article 23 versus Article 24: importer and distributor

    Where the importer’s verification duty ends and the distributor’s begins, with the locator, the evidence and the first actions per side, from the knowledge graph.

  11. Article 16 versus Article 23: provider or importer

    Whoever sources a system from outside the Union carries a verification duty, not a design duty. The two roles side by side, with the locator and the evidence per side, from the knowledge graph.

  12. Article 53 versus Article 55: GPAI with and without systemic risk

    The base set for every GPAI model provider next to the heavier package for systemic risk. Per side the locator and the evidence, from the knowledge graph.

  13. Article 12 versus Article 72: logging and post-market monitoring

    Recording logs is not the same as monitoring a system. Per side the locator, the piece of evidence and the first actions, from the knowledge graph.

  14. Article 72 versus Article 73: monitoring and incident reporting

    A continuous process next to a reporting duty with a deadline. Per side the locator, the oversight and the evidence, from the knowledge graph.

  15. Conformity assessment versus quality management system

    The continuous system next to the assessment moment, with the locator, the piece of evidence and the first actions per side, from the knowledge graph.

  16. Article 50 versus Annex III: what applies now and what gets heavier later

    A transparency duty independent of the risk level, next to the classification that opens the heavy requirement set. Per side the application date and the locator, from the knowledge graph.

  17. Article 57 versus Article 60: testing in a sandbox or in the real world

    Two routes to test before a system reaches the market, with different conditions. Per side the locator, the oversight and the evidence, from the knowledge graph.