An assessment, required by Article 27 of the EU AI Act, that public bodies and certain other deployers must carry out before using a high-risk AI system, identifying affected people, risks of harm, human oversight and complaint mechanisms.
In more detail
The assessment complements the data protection impact assessment required by the GDPR and the conformity assessment that providers of high-risk systems must complete. Deployers must notify the market surveillance authority of the results. It is one of the main ways the AI Act extends algorithmic impact assessment from voluntary practice to legal obligation. Critics worry that, as with other impact assessments, it may become a paperwork exercise without independent review.