BerriAI/litellm · error · GuardrailRaisedException

Sensitive data detected by {self.guardrail_name}

Error message

Sensitive data detected by {self.guardrail_name}

What it means

This GuardrailRaisedException is the guardrail doing its job in block mode: sensitive data matched a detection rule, the guardrail is not configured to reroute (should_route_on_sensitive_data() is false), so the call is aborted with 'Sensitive data detected by <guardrail_name>'. The exception propagates out of the guardrail hook and the request never reaches the provider unless caught or excluded.

Source

Thrown at litellm/integrations/custom_guardrail.py:431

            GuardrailRaisedException: When configured to block, or when routing is
                configured but no session_id is available
        """
        if self.should_route_on_sensitive_data():
            try:
                self.raise_sensitive_data_route_exception(
                    route_to_model=self.sensitive_data_route_to_model,
                    request_data=request_data,
                    detection_info=detection_info,
                )
            except ValueError:
                raise GuardrailRaisedException(
                    message=(
                        f"Sensitive data detected by {self.guardrail_name} (routing skipped: request has no session_id)"
                    ),
                    guardrail_name=self.guardrail_name,
                )
        else:
            raise GuardrailRaisedException(
                message=f"Sensitive data detected by {self.guardrail_name}",
                guardrail_name=self.guardrail_name,
            )

    @staticmethod
    def get_config_model() -> type["GuardrailConfigModel"] | None:
        """
        Returns the config model for the guardrail

        This is used to render the config model in the UI.
        """
        return None

    @classmethod
    def get_supported_event_hooks(cls) -> list[GuardrailEventHooks] | None:
        """
        Returns the event hooks this guardrail supports, for the UI to render.

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Remove or redact the sensitive content from the prompt before sending it
2. If masking is preferred over blocking, enable mask mode (e.g. pii_replace_with / sensitive_data_route_to_model or guardrail-specific mask config) so text is transformed instead of rejected
3. Scope the guardrail: disable it for specific keys/teams, exclude entity types, or tighten the detection rules that produced the false positive

Example fix

# before
messages = [{"role": "user", "content": "Email john.doe@corp.com about the invoice"}]

# after (pre-redact, or use a masking guardrail config)
messages = [{"role": "user", "content": "Email <EMAIL> about the invoice"}]

from litellm.integrations.custom_guardrail import GuardrailRaisedException

try:
    result = litellm.completion(model="gpt-4o", messages=msgs)
except GuardrailRaisedException as e:
    # policy decision: inform user, log, or sanitize + retry
    return sanitize_and_retry(msgs, e)

When it happens

Trigger: A guardrail without sensitive_data_route_to_model (mask/route not enabled) runs on a request whose prompt or messages contain a matched entity (e.g. Presidio PII, bedrock guardrail, regex custom guardrail) and detection returns true. Happens on pre_call/during_call hooks depending on the guardrail's event_hook config.

Common situations: PII blocking enabled company-wide and users pasting SSNs/emails/credit-card numbers; custom guardrails whose detect() is too broad (blocking benign inputs); testing guardrail configs with the same prompt used to test detection.

Related errors

• Sensitive data detected by {self.guardrail_name} (routing sk
• Cannot route sensitive data without a session_id. Ensure the
• Event hook {hook} is not in the supported event hooks {suppo
• Event hook {event_hook} is not in the supported event hooks
• Guardrail {self.guardrail_name or type(self).__name__} imple

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/566a33f675460387. Report an issue: GitHub.