Short answer
The Einstein Trust Layer is the set of security controls Salesforce places between your data and the large language models behind its generative AI and Agentforce features. It masks sensitive data in prompts, enforces zero data retention with model providers, checks outputs for toxicity, and keeps an audit trail of AI activity.
Einstein Trust Layer explained
When a Salesforce AI feature sends a prompt to a model, the Trust Layer grounds the prompt in data the user is allowed to see, can mask personally identifiable information before it leaves Salesforce, and relies on agreements with model providers that they will not retain or train on the data. Responses are scored for toxicity and the prompt, response and feedback can be logged for review.
For regulated firms, the Trust Layer is a large part of the answer to the question compliance teams ask first: where does our client data go when someone uses AI?
How Vantage Point helps: we configure masking, audit logging and permissions as part of every Agentforce rollout and document the controls for your compliance team.
Frequently asked questions
Does the Einstein Trust Layer stop models training on my data?
Salesforce's agreements with its model providers include zero data retention, meaning prompts and responses are not stored or used to train the provider's models.
Is the Trust Layer enough for compliance on its own?
It is a strong technical control, but firms still need their own AI policy, user permissions, supervision and record-keeping to meet regulatory expectations.