May 14, 2026: Cortex AI Guardrails support for Snowflake Intelligence and Cortex Agents (General availability)¶

Cortex AI Guardrails now provide run-time protection against prompt injection and jailbreak attacks for Snowflake Intelligence and Cortex Agents, in addition to Cortex Code.

Cortex AI Guardrails are part of Snowflake Horizon Catalog and use contextual reasoning to detect and neutralize adversarial threats, including indirect prompt injections embedded in tool calls. Account administrators enable guardrails for all supported surfaces with a single account-level setting using the AI_SETTINGS parameter.

For more information, see Cortex AI Guardrails.