AAIR: Advanced in AI Risk

Data handling during both model training and inference determines whether sensitive inputs or outputs may be retained, used for training, exposed, or handled in ways that violate security, privacy, or regulatory requirements.