Question

A compliance officer is concerned that the third-party large language model (LLM) used by Agentforce and generative AI features might store the company's customer prompts and use them to train the external model. She asks the administrator what protects Salesforce data from being retained by the external LLM provider. Which Einstein Trust Layer capability should the admin cite?

Accepted Answer

The zero-data-retention policy, which ensures the LLM provider does not store prompts or responses and does not use them for model training

Answer

Data masking, which permanently deletes sensitive fields from all Salesforce records before they are displayed

Answer

Dynamic grounding, which encrypts the Salesforce database so the LLM cannot read stored records

Answer

A sharing rule that restricts which users can view the LLM's generated responses