PII Detection

Agents process user data, logs, and documents containing PII. Leaking PII in responses or traces violates GDPR/HIPAA and destroys trust.

PII detection is a redaction pen that runs automatically — black out sensitive lines before they leave the building.

Scenario

User pastes SSN in chat; output guardrail redacts before display; trace stores `SSN-[REDACTED]` not the real number.

Solution

In Security & Guardrails, apply PII Detection to this scenario: User pastes SSN in chat; output guardrail redacts before display; trace stores `SSN-[REDACTED]` not the real number. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to PII Detection (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where PII Detection happens in the code.

1from presidio_analyzer import AnalyzerEngine  # import dependencies2analyzer = AnalyzerEngine()3results = analyzer.analyze(text=response, language="en",4    entities=["EMAIL_ADDRESS", "US_SSN", "CREDIT_CARD"])5redacted = anonymize(response, results)

• Treating PII Detection as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for pii detection

Quick recap — the most important points from this module.