Pre-AI Sanitization Tool

Sensitive text deserves a
clean pass before it goes anywhere

Blackout scans any text document — transcripts, reports, research notes, HR files — for names, organizations, projects, and sensitive data. Redact or code-name everything in seconds, then export a clean version you can safely share, file, or feed into AI tools.

Launch Blackout → See how it works

No account required to start · Works in your browser · Content never stored

Redaction runs locally in-browser
Content never stored on our servers
AI scan via secure backend — never logged
Export as .txt, .md, .docx, .pdf
Reversible code-name system
What it looks like

From raw and exposed to clean and shareable

Paste any sensitive document. Blackout flags every risk. Choose how to handle each one — then export exactly what you need.

① Detected
transcript.txt — Scan complete
INTERVIEWER [00:00:12] Tell me about the move from Meridian Health to Vantage Group. RESPONDENT [00:00:19] Sarah Ellison brought me in for Project Holdfast. There was a $4.2M overrun that CFO Torres didn't want disclosed.
P People
O Orgs
R Projects
S Sensitive
② Redacted
transcript_redacted.pdf
INTERVIEWER [00:00:12] Tell me about the move from ████████████████ to ████████████. RESPONDENT [00:00:19] ███████████████ brought me in for ████████████████. There was a ████████████ that ███████████ didn't want disclosed.
██ All sensitive content removed
✓ Safe to share
③ Code-named + decoder key
transcript_coded.txt
INTERVIEWER [00:00:12] Tell me about the move from ORG-1 to ORG-2. RESPONDENT [00:00:19] PERSON-A brought me in for PROJECT-1. There was a SENSITIVE-1 that PERSON-B didn't want disclosed.
CODE Readable but anonymous
decoder_key.json — store separately
"PERSON-A""Sarah Ellison"
"PERSON-B""CFO Torres"
"ORG-1""Meridian Health"
"PROJECT-1""Project Holdfast"

Four steps. Minutes, not hours.

01 — Paste

Drop in your text

Paste any text — transcripts from Otter, Zoom, or Granola; reports, research notes, HR documents, legal memos. If you can copy it, Blackout can process it.

02 — Scan

Detect sensitive terms

An instant pattern scan catches names, orgs, and keywords. Upgrade to AI Deep Scan to catch subtler references — things mentioned in passing that rules-based scanning misses.

03 — Redact

Review and act on every flag

Click any highlighted term to black it out, assign a code name, or mark it safe. Review by category — handle all people at once, or go one by one.

04 — Export

Download the clean version

Export as plain text, Markdown, Word doc, or PDF. If you used code names, download the decoder key separately and store it away from the document.

Output formats

Export exactly what you need

Every export format is designed for a real workflow — from quick file attachments to formatted client deliverables.

.txt

Plain Text

Clean, portable. Works everywhere. Great for archiving or pasting into other tools.

.md

Markdown

Speaker labels formatted as headings. Paste straight into Notion, Obsidian, or GitHub.

.docx
Pro

Word Document

Formatted for sharing with clients or filing in document management systems.

.pdf
Pro

PDF

Print-ready, professional. Ideal for legal review, compliance filing, or formal delivery.

.json
Pro

Decoder Key

A JSON file mapping every code name back to the real term. Store it securely, separately from the sanitized document.

Built for anyone who handles sensitive text

Consulting

Management Consultants

Sanitize client interview transcripts and internal reports before filing in shared drives or presenting findings to other teams.

Research

Qualitative Researchers

Anonymize participant transcripts and research notes to meet IRB requirements and protect respondent identity in publications.

Legal

Legal & Compliance Teams

Redact privileged information and sensitive figures from transcripts and memos before sharing across matters or teams.

Journalism

Journalists & Reporters

Protect source identity in working documents while keeping the full context accessible for your own review.

HR & People

HR & People Teams

Anonymize interview transcripts and investigation notes for sharing with stakeholders who don't need to know who said what.

Product & UX

Product & UX Teams

Share user research transcripts and session notes with stakeholders without exposing participant names or identifying details.

Privacy & Security

Designed from the ground up for confidential content

The tool you use to protect private documents shouldn't itself be a privacy risk.

🔒

Redaction happens in your browser

All pattern-matching and redaction runs locally in JavaScript. Your text is never transmitted to our servers for this step — when you close the tab, it's gone.

AI scan is processed securely — never stored

The optional AI scan sends your text to Blackout's secure backend, which calls Claude AI on your behalf. Your content is processed and returned — never logged or stored.

🗝️

Decoder keys are yours to store

When you use code names, the decoder key is a file you download and control. Blackout never holds it. The sanitized document is meaningless without it.

📋

Minimal account data

An account stores only your email and subscription status. No document content, no scan history, no activity logs — ever.

Simple, honest pricing

Start free. Upgrade when you need .docx, .pdf, and the decoder key.

Free
$0

3 complete exports. Full redaction, code-naming, AI scan. No credit card needed.

  • 3 free exports
  • Pattern-based scan
  • AI Deep Scan (free, sign in required)
  • Redact & code-name
  • Export .txt and coded .txt
  • .docx export
  • .pdf export
  • Decoder key download
Launch free →
FAQ

Common questions

No. For pattern-based redaction, everything runs in your browser — nothing is transmitted anywhere. For the AI scan, your text is sent to Blackout's secure backend to process the AI request, but it is never stored, logged, or retained. When you close the tab, it's gone.

The AI scan sends your text to Blackout's secure server, which passes it to Claude AI for analysis. Claude identifies names, organizations, projects, and sensitive items that pattern-matching rules miss — things mentioned in passing, indirect references, or context-dependent details. The result comes back to your browser. Nothing is stored.

Instead of blacking out a term, you can replace it with a neutral label like PERSON-A or ORG-1. This keeps the document readable while hiding the real identity. When you export, you can download a JSON decoder key that maps each code name back to the real term. Store that key separately from the document — ideally in a password manager or encrypted folder.

Not for the free tier. You can paste, scan, redact, and export .txt and .md files without creating an account. An account is required for Pro features (.docx, .pdf, decoder key download) and to track your AI scan usage.

Anything you can paste as text. Transcripts from Otter, Zoom, Granola, or Rev; reports and memos from Word or Google Docs; research notes; HR documents; legal correspondence. If you can copy the text, Blackout can process it.

Blackout uses passwordless magic link login. Enter your email, get a link, click it. No password to remember or to leak.

Yes. Cancel anytime from the Account panel inside the app. You keep Pro access until the end of the billing period.

Your documents are more sensitive than you think.

Blackout takes five minutes to learn and protects every name, number, and detail you're not ready to share — every time you use it.

Launch Blackout — it's free →

No install · No account required · Works in any modern browser