What is Privacy.ly Filter?

Privacy.ly Filter helps people find and remove likely personally identifiable information from PDFs, email HTML, and text. It suggests redactions for review, then creates a redacted copy and a metadata-only clearance packet before the document is shared.

Does Privacy.ly Filter guarantee anonymization or compliance?

No. Privacy.ly Filter provides redaction assistance and data minimization support. It requires human review and does not guarantee anonymization, GDPR compliance, or complete removal of every personal-data trace.

Which files can Privacy.ly Filter process?

Free Flex supports pasted text, HTML email samples, selectable PDFs, and coordinate-backed OCR for scanned or image-based PDFs when safe. PDFs fail closed when OCR cannot provide mapped redaction coordinates.

What happens to documents submitted through Free Flex?

Free Flex processes artifacts in request memory with a retention TTL of zero seconds, keeps provider access server-side, and uses metadata-only operational records without raw request logging.

How does Privacy.ly remove PII from a PDF?

Privacy.ly extracts selectable text or uses coordinate-backed OCR when safe, sends the extracted text through the server to a Replicate-hosted deployment based on the open-source openai/privacy-filter project, maps likely PII back to PDF locations, lets the user approve each finding, applies true PDF redaction, clears hidden document containers, and re-extracts the output to check that approved text is no longer recoverable.

PII removal for documents

Remove PII before you share.

Upload a PDF, email, or text file. Privacy.ly finds likely personally identifiable information (PII), lets you approve each redaction, and creates a redacted copy plus an evidence report.

1
Upload
PDF, email, or text
2
Find PII
Automatic scan
3
Review
Approve redactions
4
Download
Redacted Copy + evidence

Free Flex is queued. A model that has been idle can take longer to start.

Upload a document

Drop it here.

PDF, HTML email, or text. We scan for likely PII automatically.

Choose document

Paste text or email HTML to find PII

We’ll flag likely personal information for your review.

0 characters ready

Advanced settings Detection sensitivity

Status

Add a document to find PII

Privacy.ly scans it for likely personal information, then asks you to review every suggested redaction.

Output package

Clearance packets are metadata-only evidence of which PII was redacted, kept, and reviewed.

0 documents 0 ready 0 in review 0 downloadable

Documents

Select a document to review its next step.

Select all

No documents added yet.

No active document

Upload or paste a document

The review workspace appears after a document is added.

Idle

Upload a file or paste text to begin.

Preview before and after

Source Waiting for document

Add a document or use a demo-safe sample.

Reduced Copy Awaiting analysis

Reduced preview appears after analysis and review.

Review and clear for sharing

No active findings.

Show private item details

No findings yet.

How it works

How Privacy.ly removes PII from a PDF.

Privacy.ly uses the open-source openai/privacy-filter project, served through Replicate, to find likely PII. The model suggests what may be private; you decide what is actually removed.

During processing: extracted text is sent to Replicate through our server. Provider credentials stay server-side. Privacy.ly does not retain source files, filenames, or raw detected values in app-managed storage.

01
Read the PDF
Selectable text is extracted directly. Image-only pages use coordinate-backed OCR only when the words can be mapped safely back to the page; otherwise processing stops.
02
Find likely PII
The Privacy Filter model returns likely names, email addresses, phone numbers, identifiers, and other PII with a label and confidence score.
03
Review every finding
Privacy.ly maps each finding to its location in the PDF. You choose which findings to redact and which to keep.
04
Redact, clean, and verify
Approved findings are removed with true PDF redaction, not a visual cover. Metadata and embedded files are cleared, form and annotation content is baked, and the result is re-read to check that approved text is no longer recoverable.

Processing

Start free. Move faster when Priority arrives.

The review workflow stays the same. Only the time to start changes.

Free Flex

Available

Queued processing with no speed promise. An idle model can take several minutes to start.

No account
No payment
Same review controls

Priority

Coming soon

Tell us what you process. We’ll invite a small group when always-warm capacity is ready.

Email What would you process?

This is an access request, not a purchase.