Data Governance (Article 10 EU AI Act)
Training data quality obligation for high-risk AI
TL;DR
Data governance covers all practices that ensure training data quality: source assessment, annotation, cleaning, bias detection, documentation.
What is Data Governance (Article 10 EU AI Act)?
Obligations under Art. 10(2)-(5) of the EU AI Act (Regulation (EU) 2024/1689):
- Representative data selection for the intended purpose
- Examination for bias
- Minimum data quality standards
- Pseudonymisation where possible
- Documentation of data provenance
Practical example
HR AI training with 50k application records: the bias test shows an 8% preference for male applicants. Measure taken: re-balancing plus documentation.
Frequently asked questions
Special categories of data under Art. 9?
Art. 10(5) permits special categories for bias testing — subject to strict conditions.
Fines?
Art. 99 — up to EUR 15 million / 3% for infringements.
See also
Practitioner note: This article is practice-oriented compliance documentation, not legal advice. We are compliance specialists, not a law firm. For legally binding advice, consult an admitted lawyer.