Identify key legal and ethical failures in AI systems caused by unconsented, biased, or untraceable training data.
Explain how privacy regulations (e.g., GDPR, PIPA) and international standards (e.g., ISO/IEC 5259) shape requirements for dataset governance.
Distinguish between technical documentation and evaluative tools used to assess data integrity.
Apply fairness evaluation methods such as bias audits, intersectional representation scoring, and cause-effect test case generation.
Assess the audit readiness and risk profile of a real-world dataset using a structured governance framework.
Recognize early warning signs of dataset decay, including drift, annotation noise, and metadata erosion, and describe strategies for detection and mitigation.
Evaluate the performance, legal, and ethical consequences of deploying outdated or decaying datasets in high-stakes AI systems.
Propose curation and remediation strategies using technical methods such as synthetic data generation, differential privacy, and automated anomaly detection.
