← Blog  ·  April 28, 2026  ·  7 min read

The best PII redaction tools online in 2026

The landscape for PII redaction tools changed significantly in April 2026 when OpenAI released its Privacy Filter model — the first LLM-native PII detector available as a hosted API. This review covers five tools a developer or compliance team would realistically evaluate today, including both hosted services and self-hosted options.

Disclosure: We operate PrivacyFilter.run, which wraps OpenAI Privacy Filter. Our ratings are based on objective criteria; see methodology below.

1. PrivacyFilter.run

PrivacyFilter.run

Free / $9 one-time (50 redactions) / $19/month unlimited

Hosted PII redaction powered by OpenAI Privacy Filter. Paste text in the browser or call the API. Detects 10 entity types with contextual accuracy. Returns character offsets and redacted text. Three redaction modes: replace ([TYPE_N]), mask (████), tag (<PII>). Batch endpoint for paid users.

  • Zero setup, works instantly
  • Contextual name detection
  • No text stored server-side
  • Generous free tier
  • EU-friendly privacy policy
  • Requires internet connection
  • Free tier: 3/day, 2k chars
  • No custom entity types (yet)
Best for: developers, analysts, teams needing fast contextual redaction

2. Microsoft Presidio

Microsoft Presidio

Free (open source) · self-hosted

Open-source Python library for PII detection and anonymization. Built on spaCy NER + regex recognizer pipelines. Highly extensible: add custom recognizers for internal ID formats. Runs entirely on your infrastructure.

  • No data leaves your infra
  • Free at any volume
  • Custom entity recognizers
  • Active Microsoft backing
  • Complex setup (spaCy models)
  • Misses contextual PII
  • English-primary (community multilingual)
  • ~700MB Docker image
Best for: regulated industries needing on-premises, high-volume batch jobs

3. Nightfall AI

Nightfall AI

Enterprise pricing (contact sales)

DLP and PII detection platform targeting enterprises. Integrates with Slack, Google Drive, GitHub, Jira. ML-based detectors with policy enforcement and alerting. API available but pricing requires a sales conversation.

  • Deep SaaS integrations
  • Policy engine + alerting
  • SOC 2 Type II certified
  • Expensive (enterprise-only)
  • Overkill for small teams
  • Slow sales cycle
Best for: enterprises needing DLP across SaaS stack, not just text redaction

4. AWS Comprehend (PII detection)

AWS Comprehend — PII Detection

$0.001–$0.002 per unit (1 unit = 100 chars), no free tier for PII

AWS managed NLP service with a dedicated PII detection mode. Returns entity types and confidence scores. Redaction available as a batch job on S3-stored documents. Requires AWS account, IAM setup, and SDK familiarity.

  • Integrates with AWS data pipelines
  • Scales to any volume
  • HIPAA eligible
  • AWS lock-in
  • Complex pricing
  • No standalone web UI
  • Slower setup than REST API
Best for: teams already on AWS processing documents at scale

5. Google Cloud DLP

Google Cloud DLP

$1–$3 per GB inspected, free tier up to 1 GB/month

Google's Data Loss Prevention API. Supports 150+ infoTypes including jurisdiction-specific IDs (many countries). Strong for structured data (BigQuery, GCS). REST API with good Python SDK. JSON-configurable inspection templates.

  • Widest entity type catalog
  • Structured data integrations
  • 1GB/month free
  • Google cloud lock-in
  • Complex configuration for simple use cases
  • No contextual LLM-based detection
Best for: teams on GCP needing broad infoType coverage at data-warehouse scale

How to choose

Try PrivacyFilter free — no setup, no account, 3 redactions/day.

Paste text and see PII detected in seconds →

Keep reading