Security Guardrail

Redacts sensitive info (PII)

Security Guardrail

AI-generated

Summary

Redact email addresses from a given text by replacing them with placeholder tokens to hide sensitive information from AI processing, and provide a method to restore the original emails if needed.

Inputs

  • Redaction Type — Selects the kind of data to hide; currently supports only 'Email Redaction' to mask email addresses.

Output shape

A single JSON object containing methods and data related to redaction, including 'redact' for replacing emails with tokens, 'restore' for reverting tokens back to emails, and a mapping of tokens to original emails.

The output provides the redaction service with two asynchronous functions: 'redact(text)' returns the redacted text and a secret map of tokens to emails; 'restore(text, secretMap)' reverses the redactions. There is no automatic input; the user must call these methods within their workflow to process text.

Links

Discussion