OCR + anonymization · Local AI in the EU
Scanned PDF anonymizer with OCR
Scanner or photo PDFs with no text layer: OCR inside the anonymization flow to detect and redact personal data before sharing.
A scanned PDF is not selectable text—it is a stack of images. Manual blacking-out or tools that only read the text layer miss names, dates and addresses on the scan. Anonimatum runs real OCR inside anonymization so those pixels become processable text and then automatic detection targets.
Processing stays on European infrastructure. Pages are not sent to external LLM APIs: detection combines patterns, word sets, whitelist/reverse lists and local AI. Review matches before redaction and export an anonymized PDF fit for archive, publication or third-party delivery.
If you only need a searchable text layer without redaction, use searchable PDF OCR. When the goal is minimization and GDPR on a scan, this flow is the right one: OCR + review + redaction in one environment.
How to anonymize a scanned PDF
- 1
Upload the PDF or images
Drop the scan into the app. The system detects a missing text layer and prepares server-side OCR in the EU.
- 2
Review detection after OCR
Patterns, word sets and local AI flag dates, addresses, amounts, QR codes and other sensitive data. Adjust matches before redacting.
- 3
Download the anonymized PDF
Export a file ready to archive, publish or share. Your original stays under account control.
What you get
- Real OCR inside scanned PDF anonymization
- Detection with patterns, word sets, whitelist and local EU AI
- Active detectors: dates, addresses, amounts and QR
- Preview and review before redaction
- Processing on European infrastructure (GDPR)
- Batch queue compatible when plan allows
- Works alongside searchable PDF and PDF to Word OCR
- No external LLM APIs in the detection pipeline
Who this flow is for
Public bodies and archives
Digitized case files that must be published or shared without identifying data.
Law and compliance teams
Scanned contracts or opinions where a text layer is missing or incomplete.
Healthcare and insurance
Photographed reports that still contain patient or policyholder data.
GDPR owners
Teams that need evidence the scan was OCR’d and reviewed before release.
Why anonymize scans here
OCR and redaction in the same EU environment—no third-party generative LLM APIs.
OCR inside anonymization
No external pipeline: the scan becomes text and sensitive data is flagged in one process.
Local AI in the EU
Contextual detection without shipping documents to third-party model APIs.
Patterns, word sets and whitelist
Deterministic rules plus human review to reduce false positives and negatives.
Batch queue
Process volumes of scans in queue when your plan allows, without skipping review.
Common use cases
Historical files
Old digitizations without prior OCR that must enter a transparency portal.
Mobile receipts
Photos of invoices as PDF that must be forwarded redacted.
Scanned minutes
Paper-signed minutes with names and addresses to hide.
Path to Word OCR
When you also need to edit, combine with PDF to Word OCR around anonymization as needed.
FAQ — scanned PDFs
Can I anonymize a PDF without selectable text?
Yes. The flow enables OCR to extract text from image pages, then runs detection and redaction.
Does OCR run outside the EU?
No. OCR and anonymization run on Anonimatum’s European infrastructure.
What if I only need a searchable PDF?
Use the searchable PDF OCR tool when you do not need redaction.
Which AI detectors work on scans?
Active options include dates, addresses, amounts/quantities and QR codes, plus patterns and word sets.
Is manual review required?
Review is recommended. Whitelist and reverse lists help tune what is redacted vs kept.
Where do I open the flow?
At https://app.anonimatum.com in the anonymization area.