No model training
Nothing you submit is used to train or fine-tune any model.
Security, by design
Your documents.
Processed, then discarded.
PII Purge is built to hold as little data as possible. Documents are processed in memory, never stored and never used to train models.
From upload to result
What happens to your document?
Sent encrypted
Your browser or application sends your document over HTTPS.
Processed in memory
Our multi-step pipeline extracts text and layout, finds personal information in context and matches it to its position. Open-weight LLMs run on bare-metal servers in Hexploits Cloud. Nothing is sent to a third-party AI service.
Returned and discarded
The redacted copy is sent back, then the original and copy are discarded. Neither is written to disk, logged or used to train a model.
Minimal records
We record usage.
Never document content.
We keep a record of each redaction for billing, support and preventing abuse. These records are kept for a year.
For the free tool without an account, redactions are counted against a one-way hash of your IP address that changes daily. Those counts are deleted after two days.
| We record | We never record |
|---|---|
| Time, account and API key used | Document or text content |
| Size in bytes and page count | The file name |
| Number of items redacted | What the items were |
| Duration and outcome | The redacted copy |
Our infrastructure
Hosted and operated by Hexploits.
The website, API and detection models run on Hexploits Cloud. The open-weight LLMs run on our bare-metal servers.
No document retention
Models keep nothing between requests. Documents are never written to disk.
Open-weight models
We run the models ourselves and control where they run and what they keep.
Permanent PDF removal
The data beneath the bar is gone.
Rebuilt from page images
The redacted PDF is built from images of the original pages. Text beneath each redaction bar no longer exists in the output.
Hidden content removed
Original metadata, bookmarks, attachments, form fields, comments and hidden text layers are left out of the redacted PDF.
Fails closed
If detection is unavailable, the request returns an error. Scanned pages without text are rejected instead of being returned partly redacted.
Permanent removal applies to the items detected and redacted. Automated detection can miss information, so always review your result before sharing it.
Account protection
Keys, instructions and payments.
Questions about security or data protection? Get in touch with Hexploits Ltd.
Read the full privacy policy- API keys are shown once when created and stored only as a SHA-256 hash. You can revoke them at any time.
- Payments are handled by Stripe. We never see or store your card number.
- Your instructions are stored with your account so they can be applied to your redactions.
Less personal information. More peace of mind.
Share what matters.
Keep the rest private.
1 redaction a day, free. No account or card needed to get started.