Redact troubleshooting: encrypted PDFs, scans without text, filled forms, and masks that stay selectable

Common problems when redacting a PDF at /app/redact, with causes and fixes. The PDF itself never leaves your browser — only its extracted text is sent to the AI for PII detection, and the redaction itself runs entirely on your device.

"This PDF is encrypted. AttachKit can't read it"

The file needs a password to open, so the scan, OCR, and apply actions are all disabled. (A PDF that's only permission-locked — openable, but restricted — is unlocked on-device automatically; you'd see an "unlocked" notice instead of this banner.)

1. If you know the password, open Unlock, remove it, and re-drop the unlocked copy here.
2. Otherwise re-export the file without a password from its source (or open it in Preview/Acrobat with the password and save a copy).
3. Note the redacted copy is rebuilt unencrypted either way — re-protect it afterward if it still needs a password.

"This PDF has no embedded text — it looks like a scan"

The pages are images with no text layer, so there's nothing for the PII scan to read yet.

1. Pick the document's language in the dropdown next to the scan button (24 languages; English is the default).
2. Click Scan with OCR (slow). OCR runs page by page in your browser — nothing is uploaded — with a progress count and a Cancel OCR button.
3. When OCR finishes, PII detection runs on the recognized text automatically.

If you instead see "OCR didn't find any text — the page may be blank or low quality", the scan resolution is too poor to read. Re-scan the paper original at a higher resolution if you can, and double-check the language matches the document.

"This PDF has filled form fields" / the button says "Flatten form fields first"

Typed form values (like an SSN typed into a W-9) live in form widgets, not the page text, so the PII scan can't see or cover them. Redact deliberately blocks Apply redactions and download until this is resolved — otherwise you'd ship a "redacted" PDF whose fields still carry the PII.

1. Click Flatten form fields in the amber banner. The form is flattened in your browser and reloaded, so the typed values become scannable page text.
2. Run Auto-find PII again on the flattened copy.

Bulk redact skips such files entirely rather than redacting them incompletely — flatten and redact them one at a time.

Redacted text can still be selected or copied

The Forensic scrub checkbox was off when you downloaded. Without scrub, the black bar is only a visual overlay — the original characters stay in the file underneath it. This is stated in the tool, and scrub is on by default in every new session for exactly this reason.

1. Go back to your original (un-redacted) file and run the scan again.
2. Leave Forensic scrub (recommended) checked — redacted pages are flattened to images so the covered text is permanently removed.
3. Download again and verify: on a scrubbed page you shouldn't be able to select any text at all.

Don't share the earlier copy: anyone can copy the text out from under the bars.

"No PII was auto-detected" or "Automatic PII detection is unavailable"

"No PII was auto-detected" can simply mean the document is clean — but review it yourself, especially form fields and comments, before sharing. "Automatic PII detection is unavailable right now" is different: the detection service didn't run at all, and the tool shows it as a warning so it's never mistaken for a clean result. Try again later, and don't treat that state as an all-clear.

"Monthly AI limit reached" or "Rate limit exceeded"

Each scan counts as one AI action against your monthly allowance (25 per month on Free, 200 on Pro, 500 on Max — the cap applies even during the private beta). Anonymous use also has a tight hourly limit (8 scans per hour by default); signed-in accounts get far more (150 per hour).

1. For a rate limit, the message tells you how long to wait — or sign in for the higher limit.
2. For the monthly limit, upgrade for a higher allowance, or review the document and note that an already-completed scan can be saved as a draft and resumed without consuming another AI action.

"Timed out — check your connection and try again"

The detection request has a 2-minute watchdog so a dead connection can't leave the scan spinning forever. Check your connection and click Auto-find PII again. If you've just finished a long OCR run, click Save draft first — the draft preserves the OCR text, so a retry won't have to redo it.

The redacted file is bigger, or its pages aren't searchable

That's forensic scrub working as designed: each redacted page is flattened to an image, which removes the hidden text but also removes the text layer and can grow the file.

1. Run the result through Searchable to OCR a fresh text layer onto the flattened pages.
2. Run it through Compress if the size matters more than image quality.

Pages without redactions are untouched, so only the affected pages change.

The file won't drop — "AttachKit caps drops at 100 MB"

The drop zone rejects files over 100 MB to keep your browser responsive. Compress the PDF first, or split it with Pages and redact the parts separately.

Bulk redact says "No files were redacted" or lists skipped files

Bulk redact only ZIPs files that were actually processed, and it tells you why each skipped file was skipped. The usual reasons: the file is a scan with no text layer (bulk mode doesn't run OCR — open it individually and use Scan with OCR (slow)), it has filled form fields (flatten it first, see above), or PII detection was unavailable for that run. Fix the listed files and re-run, or handle them one at a time in the single-document flow.

Still stuck?

Contact support — include the exact message you saw and where the PDF came from, and we'll dig in.