Definition
The suitability of datasets for the intended purpose of an AI system. The EU AI Act requires training, validation, and test data for high-risk systems to meet quality criteria such as relevance, representativeness, completeness, and error-freeness (Article 10).