--- title: "Knowledge Base" description: "How-tos, troubleshooting, reference, and concepts for PII and PHI redaction with Philter, Phileas, Philter Desktop, and the rest of the Philterd toolkit." url: "https://philterd.ai/guides/" --- Help and how-to # Knowledge Base How-tos, troubleshooting, reference, and concepts for PII and PHI redaction with Philter, Phileas, Philter Desktop, and the rest of the Philterd toolkit. ## Browse by product - [Philter Desktop: 4 guides](https://philterd.ai/support/philter-desktop/) - [Philter: 14 guides](https://philterd.ai/support/philter/) - [Phileas: 9 guides](https://philterd.ai/support/phileas/) - [Philter AI Proxy: 2 guides](https://philterd.ai/support/philter-ai-proxy/) - [PhEye: 1 guide](https://philterd.ai/support/ph-eye/) - [PhiSQL: 1 guide](https://philterd.ai/support/phisql/) - [Phinder: 1 guide](https://philterd.ai/support/phinder/) - [Phield: 1 guide](https://philterd.ai/support/phield/) - [Philter Scope: 1 guide](https://philterd.ai/support/philter-scope/) - [Philter Workspace: 1 guide](https://philterd.ai/support/philter-workspace/) - [Philter Diffuse: 1 guide](https://philterd.ai/support/philter-diffuse/) - [Redaction Policy Editor: 1 guide](https://philterd.ai/support/redaction-policy-editor/) ## Fundamentals Concept Fundamentals ### [PII vs PHI vs NPPI: An Engineer's Guide](https://philterd.ai/guides/pii-vs-phi-vs-nppi-an-engineers-guide/) PII, PHI, and NPPI are not interchangeable. What each covers, which law governs it, how they overlap, and a decision tree for classifying your own data. Concept Fundamentals ### [What is Data Redaction? A Practical Guide](https://philterd.ai/guides/what-is-data-redaction-a-practical-guide/) Data redaction removes sensitive information from documents and datasets, but covers more techniques than most realize. A guide to strategies and trade-offs. Concept Fundamentals ### [What is NPPI? Nonpublic Personal Information, Defined](https://philterd.ai/guides/what-is-nppi-nonpublic-personal-information/) NPPI (Nonpublic Personal Information) is the financial data GLBA protects. See what counts as NPPI, what's excluded, and how it differs from PII and PHI. Concept Fundamentals ### [What is PII? A Practical Guide for Engineers and Compliance Teams](https://philterd.ai/guides/what-is-pii-a-practical-guide-for-engineers-and-compliance-teams/) PII is the term everyone uses and few define the same way. A practitioner's guide to what counts as PII, how to find it in real data, and how to handle it. ## Writing redaction policies How-to Policy authoring ### [Authoring Redaction Policies with PhiSQL](https://philterd.ai/guides/authoring-redaction-policies-with-phisql/) PhiSQL is a readable query language that compiles to a Phileas redaction policy. An introduction to the syntax and how it maps to the policy JSON. How-to Policy authoring ### [Build and test a redaction policy without writing JSON](https://philterd.ai/guides/build-and-test-a-redaction-policy-without-json/) Build a redaction policy in the Redaction Policy Editor, test it on sample text, and download JSON for Philter or Phileas without writing it by hand. How-to Policy authoring ### [Redacting National and Financial ID Numbers](https://philterd.ai/guides/redacting-national-and-financial-id-numbers/) A how-to for redacting national and financial IDs (SIN, CPF, DNI, IBAN, BIC, and more) with checksum-validated patterns that reject look-alikes. Reference Policy authoring ### [The Phileas redaction policy schema](https://philterd.ai/guides/redaction-policy-schema/) What a redaction policy is, how the JSON schema is structured, and how to use it to control exactly which PII is detected and how each type is redacted. How-to Policy authoring ### [Writing your first redaction policy](https://philterd.ai/guides/writing-your-first-redaction-policy/) A hands-on walkthrough from empty file to working redaction policy: detect an entity, apply it with Philter, change how it redacts, and handle false positives. ## Techniques How-to Detection ### [Find the PII in a Folder of Files Before You Share or Migrate It](https://philterd.ai/guides/find-pii-in-a-folder-with-phinder/) Scan a folder for PII with Phinder before you share or migrate it, and get a risk-ranked report of which files hold SSNs, email addresses, and other PII. How-to Detection ### [Measure Philter's Redaction Accuracy on Your Own Documents](https://philterd.ai/guides/measure-redaction-accuracy-with-philter-scope/) Measure PII redaction precision and recall on your own documents with Philter Scope: label a gold-standard set, audit a Philter policy, read the report. Concept Privacy engineering ### [Modern Privacy Techniques, and When to Use Each](https://philterd.ai/guides/modern-privacy-techniques-and-when-to-use-each/) A practical map of privacy techniques: redaction, pseudonymization, encryption, generalization, differential privacy, and synthetic data, and when each fits. How-to Privacy engineering ### [Publish PII counts with differential privacy instead of exact numbers](https://philterd.ai/guides/publishing-pii-counts-with-differential-privacy/) Publish PII counts with differential privacy using Philter Diffuse: choose the types, set the noise scale, and track the privacy budget across releases. Concept Detection ### [Span Disambiguation: What It Does and When to Enable It](https://philterd.ai/guides/span-disambiguation/) When two Phileas filters claim the same text with different types, span disambiguation picks one. When it helps, what it costs, and how to run it. Concept File formats ### [The Hidden Difficulties of Redacting PDF Documents](https://philterd.ai/guides/the-hidden-difficulties-of-redacting-pdf-documents/) PDFs leak redacted text in unexpected ways: invisible text layers, embedded files, and metadata. Why PDF redaction is harder than it looks, with Philter's fix. Concept Detection ### [Using an LLM or Pattern-based Rules for PII/PHI Redaction](https://philterd.ai/guides/using-an-llm-or-pattern-based-rules-for-pii-phi-redaction/) Comparing the two main approaches to redacting PII and PHI: an LLM versus pattern-based rules. How each handles accuracy, cost, and GDPR or HIPAA compliance. Concept Encryption ### [What is format-preserving encryption?](https://philterd.ai/guides/what-is-format-preserving-encryption/) Format-preserving encryption (FPE) encrypts a value so the ciphertext keeps its shape and won't break downstream systems. A guide with credit-card examples. ## AI and LLM privacy Concept AI privacy ### [PII in Vector Embeddings: A Defense Guide](https://philterd.ai/guides/pii-in-vector-embeddings-a-defense-guide/) Embeddings look like just numbers, but research shows they are partially invertible. A practical defense guide for vector stores against PII recovery attacks. Concept AI privacy ### [Prompt Engineering for Privacy: Practical Patterns for Not Leaking PII](https://philterd.ai/guides/prompt-engineering-for-privacy/) Every prompt sent to an LLM is a data egress point. Six concrete patterns for structuring prompts, redacting inputs, and scanning outputs so PII doesn't leak. Reference Compliance ### [SOC 2 and HIPAA Control Mapping for Philter AI Proxy](https://philterd.ai/guides/soc-2-and-hipaa-control-mapping-for-philter-ai-proxy/) Map Philter AI Proxy features to SOC 2 Trust Services Criteria and HIPAA Security Rule safeguards, with guidance for your own attestations. Concept Detection ### [Using an LLM or Pattern-based Rules for PII/PHI Redaction](https://philterd.ai/guides/using-an-llm-or-pattern-based-rules-for-pii-phi-redaction/) Comparing the two main approaches to redacting PII and PHI: an LLM versus pattern-based rules. How each handles accuracy, cost, and GDPR or HIPAA compliance. ## Self-hosting and on-device Troubleshooting On-device ### [Philter Desktop Blocked by Smart App Control or App Control for Business](https://philterd.ai/guides/philter-desktop-blocked-by-smart-app-control/) Fix "An application control policy has blocked this file" when starting Philter Desktop. Update to 1.3.1, which signs every file Smart App Control checks. How-to On-device ### [Philter Desktop Quick Start, from Install to First Redacted Document](https://philterd.ai/guides/philter-desktop-quick-start/) Install Philter Desktop on Windows, set up a redaction policy, redact your first document, and review the draft before you share it. Nothing gets uploaded. How-to On-device ### [Redact a Folder of Documents Automatically on Windows](https://philterd.ai/guides/redact-a-folder-of-documents-automatically/) Point Philter Desktop at a folder and new files get redacted as they arrive. Watched folders, a command line, and a right-click action, all running on your PC. Troubleshooting On-device ### [Running Philter Desktop on an Apple Silicon Mac with Parallels](https://philterd.ai/guides/running-philter-desktop-on-a-mac-with-parallels/) Run Philter Desktop in Windows 11 on ARM under Parallels on an Apple Silicon Mac. Install the Visual C++ ARM64 runtime first so name detection loads. Reference Self-hosting ### [Self-Hosted Philter Reference Architectures](https://philterd.ai/guides/self-hosted-philter-reference-architectures/) Reference architectures for running self-hosted Philter inside your own AWS, Azure, or GCP boundary, with no-egress and government-cloud variants. Concept Self-hosting ### [Self-Hosted PII Detection on Your Own Hardware](https://philterd.ai/guides/self-hosted-pii-detection-on-your-own-hardware/) Run PII detection entirely inside your own boundary: serve the models with PhEye, embed Phileas, or run local ONNX inference, with no data egress. ## Integrations How-to Integrations ### [Apache NiFi and Philter](https://philterd.ai/guides/apache-nifi-and-philter/) How to use Philter to redact PII and PHI inside an Apache NiFi data flow, either through Philter's API or with an embedded NiFi processor. How-to Integrations ### [Redacting PII in Apache Kafka](https://philterd.ai/guides/redacting-pii-in-apache-kafka/) Redact PII and PHI in Apache Kafka with the philter-kafka-connect Single Message Transform, on source, sink, or MirrorMaker connectors. How-to Integrations ### [Redacting PII in Logs with log4j2 and logback](https://philterd.ai/guides/redacting-pii-in-logs-with-log4j2-and-logback/) How to redact PII from Java logs before they are written, with a Phileas-backed log4j2 rewrite policy and a logback converter, recursion guard included. How-to Integrations ### [Redacting PII in Spark and Databricks](https://philterd.ai/guides/redacting-pii-in-spark-and-databricks/) Redact PII column by column in Apache Spark and Databricks with a Phileas UDF: a verified PySpark example, plus the performance tradeoffs to plan for. How-to Integrations ### [Redacting PII in Trino with the Phileas Connector](https://philterd.ai/guides/redacting-pii-in-trino-with-the-phileas-connector/) How to install the open source Phileas connector for Trino, configure a redaction policy, and mask PII column by column in SQL, with a runnable demo. How-to Integrations ### [Redacting Text in Amazon Kinesis Data Firehose](https://philterd.ai/guides/redacting-text-in-amazon-kinesis-data-firehose/) Amazon Kinesis Firehose is a managed streaming service that moves data from sources to destinations like S3 and Redshift. This post redacts PII in that stream. How-to Integrations ### [Using Philter with Microsoft Power Automate](https://philterd.ai/guides/using-philter-with-microsoft-power-automate/) How to call Philter from a Microsoft Power Automate (Flow) automation to redact PII and PHI from text, using a simple HTTP action. ## Deployment and operations How-to Operations ### [Alerting on PII Spikes with Phield and Slack](https://philterd.ai/guides/alert-on-pii-spikes-with-phield/) Alert in Slack when PII counts spike with Phield: send per-batch counts from your pipeline, read the breach alerts, and tune the threshold with replay. How-to Operations ### [Deploying Philter in AWS via a CloudFormation Template](https://philterd.ai/guides/deploying-philter-with-aws-cloudformation/) How to deploy Philter in AWS with a CloudFormation template: finding the Philter AMI, editing the template, and launching the stack. How-to Operations ### [How to Use a Signed SSL Certificate with Philter](https://philterd.ai/guides/using-a-signed-ssl-certificate-with-philter/) How to replace Philter's self-signed certificate with one signed by a trusted authority, using a PKCS12 keystore and two settings. How-to Operations ### [How to Use an Apache Reverse Proxy with Philter](https://philterd.ai/guides/using-an-apache-reverse-proxy-with-philter/) How to run an Apache reverse proxy in front of Philter for SSL termination, access control, and access logging. How-to Operations ### [Monitoring Philter in AWS](https://philterd.ai/guides/monitoring-philter-in-aws/) How to monitor Philter in AWS: CloudWatch Logs for its log, load balancer health checks for availability, and its Prometheus metrics. How-to Operations ### [Referential Integrity with Valkey](https://philterd.ai/guides/referential-integrity-with-valkey/) How to configure a Valkey cache so Philter maintains referential integrity (consistent replacement values) across documents and contexts in a cluster. How-to Operations ### [Review Redactions Before Release: Thresholds, Sampling, and Sign-off in Philter Workspace](https://philterd.ai/guides/philter-workspace-review-thresholds-and-sign-off/) Set the PII threshold, audit sampling rate, and approval rules in Philter Workspace to decide which redactions a person reviews and who signs off. ## Compliance How-to Compliance ### [Deploying Philter in a HIPAA Environment](https://philterd.ai/guides/deploying-philter-in-a-hipaa-environment/) How to configure a Philter deployment for HIPAA: encryption of data at rest and in motion across AWS, Azure, and Google Cloud. Concept Compliance ### [PIPEDA and Law 25: PII Redaction for Canadian Privacy Compliance](https://philterd.ai/guides/pipeda-and-law-25-pii-redaction-canadian-privacy-compliance/) How Philter addresses PIPEDA and Quebec's Law 25: self-hosted deployment for data residency, bilingual redaction, and Canadian identifier handling. How-to Policy authoring ### [Redacting National and Financial ID Numbers](https://philterd.ai/guides/redacting-national-and-financial-id-numbers/) A how-to for redacting national and financial IDs (SIN, CPF, DNI, IBAN, BIC, and more) with checksum-validated patterns that reject look-alikes. Reference Compliance ### [SOC 2 and HIPAA Control Mapping for Philter AI Proxy](https://philterd.ai/guides/soc-2-and-hipaa-control-mapping-for-philter-ai-proxy/) Map Philter AI Proxy features to SOC 2 Trust Services Criteria and HIPAA Security Rule safeguards, with guidance for your own attestations. Concept Compliance ### [Why Philterd Doesn't Need to Sign a BAA](https://philterd.ai/guides/why-philterd-doesnt-sign-a-baa/) Philterd's software runs inside your own infrastructure and sends your PHI nowhere, so for the software there is no business associate and no BAA to sign.