What Sensitive Info Detection finds
Identify and handle PII, credentials, and personal data.
Sensitive Info Detection can detect customer data, personal identifiers, secrets, tokens, credentials, and other sensitive content before the model call runs.
Flag, redact, or block
Policy can flag the request for review, redact the sensitive value, or block the call entirely.
Like the other guardrails, the workspace default applies to all instances, and an instance can override it for stricter project-specific handling.
Next step
Turn this concept into a visible workflow with teams, approvals, grounding, and handoff.
Read Guardrails overview