Which feature of the Einstein Trust Layer ensures that AI models only access the data necessary for a particular task and no more?

The feature of the Einstein Trust Layer that ensures AI models only access the data necessary for a particular task is Purpose Limitation. This principle is designed to restrict the use of data to the specific purposes for which it was collected. By implementing Purpose Limitation, Salesforce ensures that AI models cannot utilize data beyond what is needed for their designated functions, thereby enhancing data privacy and security. This aligns with best practices in data governance and compliance, protecting user information from unnecessary exposure or misuse.

Data Masking, while important for protecting sensitive data, focuses on concealing specific data elements rather than limiting the access of models to only what is essential for their tasks. Zero Retention pertains to not storing data after its intended use has been fulfilled, which is not directly related to ensuring models only access necessary data. Auditing involves monitoring and reviewing data access and usage but does not dictate the limitation of access based on purpose.