Fundamental Rights Impact Assessments under the EU AI Act Who, What, and How
summary
Panel proposes a concrete FRIA method combining a questionnaire + matrix: a 6‑section multiple‑choice questionnaire yields a single risk indicator, and a 50‑line matrix (one score per Charter right) yields impact significance; formula weights questionnaire 30% and matrix 70% to produce an F‑score. Blocking responses force redesign (e.g., fully automated decisions are suspended). Thresholds: <50 pass, 50–75 needs mitigation, >75 stops deployment. Tools include an online prototype and a credit‑scoring use case; deployers include public authorities, schools, hospitals and banks.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 5:02EU AI Act obligations and FRIA roles
- 13:37Transatlantic critique of FRIA limits
- 23:32Bon University FRIA model
- 32:46Tesla’s FRIA risk tools
- 42:41FRIA implementation challenges
- 54:58Q&A
- 1:06:46Final reflections and conclusions
related talks
01h 15mFrom legal obligation to practice: making Fundamental Rights Impact Assessments work.
01h 17mASSESSING THE IMPACT ON FUNDAMENTAL RIGHTS IN AI APPLICATIONS
01h 18mWhich Impact Assessment for AI, Beyond Data Protection
01h 06mRisking Fundamental Rights The Myriad Ways of Risk in EU Law and Policy
01h 20mFLOPs and beyond Decoding the AI Act’s Systemic Risk Criteria
01h 16mBEYOND ETHICS WASHING:IMPACT ASSESSMENTS, AUDITS, AND OVERSIGHT FOR AI
01h 15mTowards agreed practices for privacy impact assessments
28mAI Act - What is important to know from a developers perspective?