Redacting PII and Sensitive Data with DSPy

Strip personal information and sensitive data from text before it reaches an LM — or before it leaves your system.

Step 1 - Understand What to Redact

Before writing code, answer three questions:

1. What PII types? Names, emails, phones, SSNs, credit cards, addresses, dates of birth, IP addresses, medical record numbers (MRNs), or all of the above.
2. Replacement strategy? See the table in Step 3.
3. Compliance requirement? GDPR (EU personal data), HIPAA (US health data), PCI-DSS (payment data), or internal policy.

The answers drive which pipeline path you need.

Step 2 - Set Up DSPy