Blog
PII & PHI Redaction Guides and Best Practices
Practical guides on PII and PHI redaction: self-hosted pipelines, LLM and RAG privacy, HIPAA and GDPR, and de-identification, from the Philter maintainers.
Start here
How to Redact PII Before Sending to an LLM: Chat, RAG, and AI Agents
Redact PII and PHI before any prompt reaches an LLM, keeping data on your own infrastructure. A self-hosted pipeline for chat, RAG, and AI agents.
How to Deploy Philter in 5 Minutes on the AWS Marketplace
Subscribe, launch into your own VPC, and redact PII in minutes. A step-by-step guide to deploying the self-hosted Philter API from the AWS Marketplace.
Automating HIPAA Safe Harbor: A Blueprint for Healthcare Data Pipelines
How the Philterd suite maps to the 18 HIPAA Safe Harbor identifiers, with a deployment blueprint for patient data lakes, research pipelines, and medical RAG.
Latest posts
Air-Gapped PII Redaction for Government: A Deployment Walkthrough
A walkthrough for standing up Philter, PhEye, and MongoDB in a disconnected enclave: pinned images, offline transfer, and proving there is no egress.
Five Ways a Healthcare App Leaks PHI Without Anyone Noticing
The demo looks clean. Here are five places PHI quietly ends up in a fast-built healthcare app, why each one happens, and what to do about it.
Redact PDFs Offline on Windows so No Data Leaves Your PC
Philter Desktop redacts PDFs entirely on your Windows PC, including scanned pages read with on-device OCR. No uploads, no server, and it works offline.
The Clean Room for AI Training Data: Redact, Review, Measure
Once PII is baked into model weights it is extractable for the model's life. The clean room pattern: Philter redacts, Arbiter reviews, Philter Scope measures.
When a PDF Is Too Dense to Redact
Redacting a 400-page file with PII on every line leaves a compliant document nobody can use. Distill it into the facts downstream needs, then redact those.
Moving PII Redaction Off the CPU: An FPGA Experiment
I put a streaming SSN redaction engine on an FPGA. It sits inline on the wire, so bytes get redacted before they reach a CPU, host memory, or an OS.
Why Your LLM Provider's Built-In Guardrails Are Not Enough
OpenAI, Anthropic, Google, and AWS offer PII filtering, but it all runs after your data leaves your network. Why filtering out is not the same as never sending.
Redact Both Sides of the LLM: Inputs and Outputs
Prompt redaction protects the provider boundary. Output scanning protects everything downstream. An LLM needs PII redaction on both sides, not one.
Can an LLM Leak Its Training Data, and Why You Cannot Un-Train PII
Research shows LLMs memorize and leak training data, and unlearning it afterward is unreliable. The dependable control is to redact before training.
Earlier posts
2026
- Designing a HIPAA Architecture on AWS
- Introducing Philter Router: The Right Policy for Every File
- Introducing Philter Desktop: Redact Documents Right on Your Windows PC
- PII Redaction Is a Lifecycle, Not a One-Time Fix
- Our Project Issues Are Now Public
- The Phileas Trino Connector Is Now Open Source
- How to Tell if a PII Redaction Model Is Any Good
- Precision, Recall, or F1: Which Redaction Metric Matters Most in Your Industry
- Redact PII From Your AI Agent in One Block of Config
- How to Deploy Philter in 5 Minutes on the AWS Marketplace
- How We Built PhEye's PII Name Models
- How We Use Philter Scope to Evaluate Our PII Models
- Multilingual Medical PII Redaction: English and French with Philter
- Cutting False Positives on National IDs with Checksum Validation
- philter-sdk-java 2.0.0: Built for Philter 4.0.0
- Phileas 4.0.0
- How to Redact PII Before Sending to an LLM: Chat, RAG, and AI Agents
- Introducing PhiSQL: The Query Language for PII Operations
- PhEye Update: Unified Branch, GPU Support, and Streamlined Testing
- Introducing Arbiter: Human-in-the-Loop PII Redaction
- The TCO of "Free" Cloud PII Redaction: AWS Comprehend, Google DLP, vs Self-Hosted at Scale
- PII Masking and Redaction in Trino with the Phileas Connector
- Redaction for Legal and E-Discovery: Privilege, Rule 9037, and the In-House Counsel's Pipeline
- The Ethics of Training: Why We Use Synthetic Data
- Building a HIPAA-Compliant Medical Chatbot
- Building a Privacy-Aware RAG System
- Redaction for Financial Services: PCI DSS, GLBA, and the Real-World Data Pipeline
- Architecting Privacy in Kafka: Real-Time Redaction for Streaming Data
- Beyond Regex: Why General LLMs Fail at PII Discovery
- Compliance as Code: Integrating Philter into Your CI/CD Pipeline
- Migrating from AWS Comprehend to Philter: A Practical Transition Guide
- Redaction for Insurance: Claims, Customer Data, and the State-by-State Patchwork
- Open Source vs. Black Box: Why You Can't Afford "Trust Me" Privacy
- Automating HIPAA Safe Harbor: A Blueprint for Healthcare Data Pipelines
- Privacy Shouldn't Be a Guessing Game: Evaluating Redaction with Philter Scope
- Why API-Based Redaction is a Security Antipattern
- Redaction for Government and Federal Workloads: FedRAMP, CMMC, ITAR, and the Air-Gap Imperative
2025
- Deploying Philter in Air-Gapped Environments
- From Phileas to Philter: The Evolution of Our Open Source Engine
- Redaction for Education: FERPA, Student Records, and Research Data Pipelines
- Snowflake PII Redaction: A Practical Integration Guide
- Redacting PII and PHI in OpenSearch Data Prepper Pipelines
- Philter 3.1.0
- Phileas 2.12.0
- Why Using an LLM to Redact PII and PHI is a Bad Idea
- Shielding Your Search: Redacting PII and PHI in Elasticsearch with the Search Redact Plugin
- Shielding Your Search: Redacting PII and PHI in OpenSearch with the Search Redact Plugin
- Phileas 2.10.0
2024
- Phileas in Graylog – Removing PII from Logs
- Phileas 2.9.1
- Automatically Redacting PII and PHI from Files in Amazon S3 using Amazon Macie and Philter
- Philter as an AI Policy Layer
2023
- Phileas: The Open Source PII and PHI redaction engine