Opaque overlay
The page looks covered, but text extraction still recovers both synthetic canaries.
Download overlay demoSHA-256
4910ADA866424F3CC9C738247710527D349A5D135DE3B42B33FB15E02C432CE0Synthetic before-and-after files
PDF redaction demo: visual covering vs verified absence
A two-file exercise showing why a black rectangle is not proof that underlying PDF text is gone—and why absence must be verified separately.
The comparison
Same visual idea, different underlying data.
The first PDF places an opaque rectangle over two synthetic canaries. They are hidden from view but remain extractable. The second PDF was authored without those canaries in its content data.
Comparison file
The two canaries were never written to this PDF and are absent from its page content and decoded streams.
Download comparison fileSHA-256
E6B3EF38247808599B8FF713CE4E1F569D09EE33DB761B26F046E8E577937A0AVerify it yourself
Do more than look at the page.
- Open both PDFs.They should render as one-page synthetic documents.
- Search the files.Search near the covered region or use a PDF text extraction tool.
- Try selection and copy.Visual covering should not be trusted when underlying text can still be selected or extracted.
- Inspect hidden structures.A real workflow must also consider metadata, annotations, attachments, forms, layers, OCR, and other objects.
- Review the meaning.Absence of these two canaries does not prove that every sensitive fact is gone.
Next step
Apply the full upload decision.
Check destination approval, data minimisation, hidden content, output usability, and human review.