Article 10 of the EU AI Act: Data & Data Governance
Article 10 is where fairness is engineered in: high-risk AI must be built on data that is relevant, representative and examined for bias.
What Article 10 requires
Training, validation and test datasets must meet quality criteria: appropriate data-governance practices covering design choices, collection and provenance, preparation, assumptions, prior assessment of availability and suitability, and examination for possible biases likely to affect health, safety or fundamental rights — plus measures to detect, prevent and mitigate them.
Who it binds
Providers of high-risk AI systems. Data governance sits alongside the risk-management system (Article 9).
Key points
- Data must be relevant, sufficiently representative and, as far as possible, error-free.
- Bias must be actively examined for and mitigated.
- Covers provenance and the full data lifecycle, not just the final dataset.
FAQ
Does Article 10 require perfect data?
No — it requires data that is relevant, sufficiently representative, and examined for bias to the best extent possible, with governance over the data lifecycle.
Related articles
Run the free 2-minute checker — deterministic, no sign-up.