databricks Databricks Release Notes ·

Databricks enhances guardrails with sensitive data detection and expanded capabilities

securitydatabrickspreviewengineer
feature announcement

Databricks has released a new Sensitive Data Detection guardrail (Beta) that automatically identifies and blocks or redacts structured sensitive data like Social Security and credit card numbers. This deterministic detection adds minimal latency and improves data privacy and compliance. The release also includes enhancements to existing guardrails, such as standard HTTP 200 responses for blocked interactions and expanded evaluation scope for LLM-as-a-judge guardrails. To utilize these features, users must enable the Enhanced Unity Gateway preview.

  • Add new guardrail execution flow view
  • Return standard HTTP 200 for blocked interactions
  • LLM-as-a-judge guardrails evaluate broader conversation history
Features (1)
  • Add new guardrail execution flow view

    A new execution flow view is available, which displays the order in which different guardrails run from end to end. This provides greater visibility into the guardrail processing sequence.

Enhancements (2)
  • Return standard HTTP 200 for blocked interactions

    Blocked interactions now issue a standard HTTP 200 response with the block reason instead of a 4xx error. This prevents coding agents from re-triggering the same block when resending conversation history.

  • LLM-as-a-judge guardrails evaluate broader conversation history

    Built-in LLM-as-a-judge guardrails can now evaluate a window of recent conversation turns or the entire conversation history. This expands their assessment beyond only the latest message, providing more comprehensive evaluation.

Read the original announcement →

https://docs.databricks.com/aws/en/release-notes/unity-gateway#block-or-redact-sensitive-data-in-requests-and-responses

Related releases