Snowflake Cortex AI
Snowflake says Usage and Customer Data for Snowflake AI Features, including inputs and outputs, are not used to train, re-train, or fine-tune models made available to other customers and remain inside the Snowflake security boundary.
Customer inputs and outputs are not used to train models offered to other customers. Account-specific fine-tuned models stay in that account unless the customer directs otherwise.
No opt-out is required for the stated training baseline. Teams can restrict access with role-based controls and model allow-lists.
Snowflake states that inputs and outputs remain within the Snowflake Security Boundary and are not available to other customers.
Observability and agent-monitoring features can store traces and events in account-local tables; those paths require separate review even when model training is restricted.
The Trust and Safety FAQ focuses on training, isolation, and governance rather than one universal human-review statement.
Foundational reference pages for teams adopting, deploying, and operating production AI with inspectable controls.