Safety & Security · Software component
Document PII Redactor
Software componentSafety & SecuritySafety, Security & Governancearc:DocumentPIIRedactor
An ETL transformation component that replaces detected PII spans in source documents with typed placeholder masks before chunking and embedding, routing uncertain detections to human review.
Responsibility. Prevents PII from entering the knowledge base.
Also known as: Ingestion-time PII redaction, PII masking, PII redaction stage
Relationships
invokes dependency
- NER PII Detector Ch7.5
- PII Detector abstract Ch6.4
writes dependency
emits telemetry to dynamic
escalates to dynamic
receives data from dynamic
sends data to dynamic
- Document Chunker abstract Ch6.4
guards control
- Vector Index Store abstract Ch6.4
is orchestrated by control
Design guidance
- MUST redact PII during ETL before chunking and embedding, since there is no 'un-embedding' operation short of complete re-ingestion.
- SHOULD auto-redact high-confidence (>=0.90) detections and flag medium-confidence (0.70-0.89) detections for human review.
- SHOULD write PII detection audit logs for compliance reporting.
- SHOULD replace PII with typed placeholders (e.g., [PERSON], [EMAIL], [PHONE]) rather than deleting it, to preserve sentence coherence for training.
Quantitative guidance
As stated by the sources; verify before use.
- Missing ingestion PII detection led to breach notification and $1.2 million in fines (Ch6.4 case study).
- 7% of one healthcare knowledge base contained patient PII (Ch6.4 case study).
Classification
- Patterns
- Typed placeholder maskingConfidence-based risk filteringRedact-before-embed
- Technologies
- PresidioNVIDIA NeMo Curator
- Quality attributes
- Privacy (NIST AI RMF: privacy-enhanced)Transparency and accountability (NIST AI RMF: accountable and transparent)
- Risks mitigated
- PII becoming permanently retrievable once embeddedHIPAA breach via RAG responsesModel memorizing and regurgitating training-data PII
- Frameworks & regulations
- HIPAA
Sources
- Ch6.4: T. Nguyen, "Data Quality Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 6.4. ISBN: 9798244538229.
- Ch7.5: T. Nguyen, "NeMo Curator, Riva Speech AI & Multimodal Integration," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 7.5. ISBN: 9798244538229.