rego-art10-001

rules/rego/complyedge/highrisk/art10_data_governance.rego

Citation

Regulation (EU) 2024/1689, Article 10(2) and (3)

Condition

Training, validation or test data for a high-risk AI system that is knowingly biased, unrepresentative, incomplete or unexamined, or an instruction to train or ship on such data without the data governance Article 10 requires.

Layer 1 detects

"training data" directly followed by "not representative", "biased", "incomplete" or "unvetted"; "no", "without" or "inadequate", then "data governance", then "for" or "on", then "high-risk" or "training data"; and "unrepresentative" or "biased", "training" optional, then "data", "dataset" or "sample", in any word ending ("biased datasets"). It matches wording, not meaning: it misses the practice told in other words ("We know the loan dataset barely includes women applicants, but train the credit model on it anyway"), it does not read whether the system is high-risk, and it fires on text that only names the practice, such as a plan to test for biased data. Whether a text does the practice is judged in Layer 2.

Condition type

deterministic

Status in the file

approved