Redacting a scanned document means permanently removing sensitive information before sharing it — and doing it properly requires more care than it first appears.
The mistake: just drawing a black box
If a PDF has a real embedded text layer (from OCR) underneath a visual scan, simply drawing an opaque box over a section on top doesn't remove the underlying text — copying and pasting from the 'covered' area can still reveal it.
What proper redaction requires
Genuine redaction needs to either remove the underlying image data in that region entirely, or ensure no separate searchable text layer exists behind the visual cover — an editing tool built specifically for scans (that edits the actual image, not just an overlay) handles this correctly.
When this matters most
Any document being shared outside a fully trusted context — a redacted ID for a form submission, a contract shared with a third party — deserves this level of care, since a failed redaction defeats the entire purpose.
Is it safe to redact by editing the image itself rather than adding an overlay?
Yes — a tool that actually modifies the pixel data of the scanned image (rather than layering something on top) is the reliable approach, since it removes the original content rather than hiding it.