Last updated:
A self-determination mechanism under the EU AI Act allowing providers to classify an Annex III AI system as not high-risk if it poses no significant harm.
This exception is narrow. It applies only if the system performs a purely preparatory or procedural task, or improves a result without influencing the final decision. Providers must document this reasoning and notify national authorities before placing the system on the market. If the determination is found to be groundless during an audit, the organization faces full high-risk liability and potential fines.
Real world example:
A university uses AI to reformat student transcripts into a standard layout for human review. They invoke the EU AI Act Article 6(3) exception because the AI is performing a preparatory task and not making the actual admission decision.
