PII Redaction + Guardrails

The legal/security shape — Presidio-based PII detection at the gateway, prompt-injection detection inbound, content filtering outbound, and per-key guardrail policies.

💡

Tip: Guardrails belong at the gateway, not in every application. One enforcement point, one audit trail, one place to update policy.

What this use case covers

  • PII detection and redaction with Microsoft Presidio at the gateway boundary — emails, phone numbers, SSNs, credit cards.
  • Prompt-injection detection on inbound requests, before they reach the model.
  • Content filtering on outbound responses, with a configurable policy.
  • Per-key guardrail policies — some teams need stricter rules (regulated workloads) than others (internal tools).
  • A brief compliance-posture note covering SOC2, HIPAA, and GDPR concerns and where the gateway helps vs where it does not.

Why it matters

The fastest way to fail a security review is to admit that PII leaves the network in plaintext, or that there is no centralized policy on what users can ask a model. Guardrails at the gateway are the cleanest answer to both.

Coming soon

Implementation coming soon. This use case will ship as a runnable folder in infinite-containers/lite-llm, with a step-by-step lesson here and a companion post and video. Track the full publishing roadmap in USE-CASES.md.