If you need to redact document online, use a private browser workflow that lets you choose or paste a scan or screenshot, detect sensitive text, check teh boxes, and export a redacted copy.
Redact document online with AI Document Redactor. If your source is a JPG. PNG or WebP image. Plus, you want the page’s stated privacy approach.
100% private, your data never leaves your device.
The output is a permanently redacted image. That matters, because a real redaction replaces the pixels where the private text appears. It isnt the same as placing a translucent highlighter. Emoji or shape over a screenshot. In a chat app and hoping no one can recover what sits underneath. Simple as that.
This workflow is for document images. It doesnt accept PDF files, Word documents, spreadsheets or editable text files as input.
What AI Document Redactor does with scans and screenshots
AI Document Redactor lets you choose or paste a scan or screenshot of a document. Then uses on device OCR with tesseract.js to read text in the image. It’s built for people who need to hide names and emails in documents. Cover phone numbers, redact ID numbers from image files. Or redact card numbers in screenshot captures before sharing. You get the idea.
The automatic detection covers emails, phone numbers, card numbers, SSN or national ID patterns. IBANs and dates of birth in free use. The tool’s short description also says it can auto detect names, emails, IDs and card numbers via OCR. Treat that as a starting point, not a promise that every private detail will be found.
If you need to redact screenshot online, the paste flow is convenient.
If you need to redact scanned document pages, image quality matters more. A sharp scan with clear contrast gives the OCR a better chance then a tilted, blurry photo.
Redact document online with a private browser workflow
A private document redactor is useful when the file contains personal data that you dont want to send. Throught another serivce just to cover it up. The app states. Your document never leaves your browser.
Thing is, it also states that the exported image has the sensitive text permanently burned out.
Here’s how I’d choose the right approach for common jobs:
| Situation | Best approach | Watch for |
|---|---|---|
| A screenshot with an email, phone number or card number | Run "Find sensitive data", then review every detected box. | Automatic detection can miss text, so inspect the preview. |
| A scan with names, IDs and dates | Use OCR detection, then draw manual boxes over anything missed. | Low contrast or blur may produce no readable text. |
| A specific name or project term | Use Pro "Custom terms, one per line" or draw boxes manually. | Custom terms are not included in free use. |
| Many images from the same task | Use Pro batch processing for up to 50 images in one run. | Free use is one image at a time. |
| An editable PDF, Word file or spreadsheet | Use a different redaction workflow made for that file type. | This tool exports redacted images, not edited originals. |
How automatic OCR detection finds sensitive text
OCR first tries to read the text that appears in the image. Detection then looks for sensitive patterns such as emails, phone numbers, card numbers, SSN or national ID patterns, IBANs and dates of birth. Now, that pattern based approach is helpful for common formats. But it can fail on unusual spacing, partial numbers, handwriting, glare, rotation or text that’s too small. It adds up.
The UI has a “Language” control. With English, Spanish, French, German, Italian, Portuguese and Dutch.
Free OCR is English.
Pro adds OCR in 100+ languages using tesseract.js language packs. Including multi language selection.
For names or internal terms that don’t follow a standard pattern. Pro adds “Custom terms. One per line” for up to 500 names or words. Plus “Custom regex” for custom pattern matching.
If a regex is invalid, it has to be fixed or cleared. Before detection can run properly.
Step-by-step: redact a document image
- Open the private browser redactor.
- Use “Drop a image here, click to choose. Or paste a screenshot.”.
- Use “Change” if you picked the wrong image or need to replace it.
- Choose “Language” from English, Spanish, French, German, Italian, Portuguese or Dutch.
- Choose “Redaction colour” as Black. White is available with Pro.
- If you have Pro and need specific words found, add them under “Custom terms, one per line”.
- If you have Pro and need pattern matching, enter it under “Custom regex”.
- Click “Find sensitive data” to run detection.
- Review “Detected items” and toggle boxes you want to keep or remove.
- Drag on the image to draw a box manually over anything the OCR missed.
- Click a box to toggle it. To remove a manual box, select it and press Delete.
- Use “Undo” or “Redo” if you make a mistake.
- Click “Download redacted PNG” to export the redacted image. With Pro batch output, use “Download all as ZIP”.
A paid Pro plan adds more runs plus this tool’s Pro extras.
Batch processing, more languages, full resolution output. JPG and PDF output, ZIP download, white redaction, custom terms and custom regex.
Review boxes, draw manual redactions, and export safely
The review step is where redaction succeeds or fails. Detected items appear as boxes and you can toggle them before export. If an email in the header is found. But the same email in a footer is missed. You stil need to draw the secound box yourself.
Manual boxes are also the right tool. For logos, signatures, account snippets, QR like areas, handwritten notes, or labels that OCR cant read. If your trying to redact sensitive information. In image files, don’t rely only on text detection.
Look at the whole image. Including margins and thumbnails inside the screenshot. Realy.
Free redaction colour is Black. Pro adds White, which can be better. When the page background is white. Also, you want the output to look less visually heavy.
Either way, the point is the same. The exported image should show a solid redaction where the sensitive pixels used to be.
Free document redaction tool limits, formats, and export options
Free use supports one image at a time. Input file types are JPG, PNG and WebP. With a file size limit up to 15 MB.
Files larger than that are rejected. Plus the app may aslo reject images that are to large. For the device to proccess.
Free output is PNG.
Plus the free PNG export is downscaled to a maximum of 3 megapixels. Described as about 2000×1500. And free exports are clean with no watermark. And that matters.
Full resolution export isn’t part of free use. Seriously.
If your image is in a different format, convert a copy. Before redaction with the Private Image Converter. Don’t convert your only original. If you still need it unchanged for your records.
What Pro adds for batch work, languages, and full-size output
Pro is mainly for higher volume or less standard jobs.
It adds batch processing for up to 50 images. In one run and “Download all as ZIP” for batch output. Thats the difference bewteen processing one screenshot and handling a folder of scanned pages. That’s it.
It also adds full resolution export up to 40 MP. JPG output and PDF output. PDF output doesn’t mean PDF input.
The source still needs to be a supported image file.
For language and matching, Pro adds OCR in 100+ languages usign tesseract.js language packs. Including multi langauge selection, it also adds custom redaction terms for up to 500 names or words. Plus custom regex for patterns your job needs.
Important checks before sharing a redacted image
- Zoom in before exporting and check every corner of the image.
- Look for repeated values. Like the same email in a header and footer.
- Check small text, stamps, signatures, filenames, browser tabs and adress bars.
- Make sure you added at least one redaction area before exporting.
- Open the exported PNG and inspect that copy, not just the preview.
- Don’t share the original file by accident.
Redaction covers visible pixels in the exported image. If your sharing risk includes image metadata, handle that as a seperate step wiht the photo metadata cleaner. If the document is for a court filing, medical disclosure, tax matter or regulated buisness process, use this as practical workflow information and follow the rules that apply to your situation.
Troubleshooting common redaction mistakes
No readable text was found
Use a sharper, higher contrast image if you can. If the scan is still not readable to OCR, draw manual boxes over the sensitive areas.
No sensitive patterns were detected
This can happen even when the image contains private text. Review the image yourself, then draw boxes over names, emails, IDs, dates or numbers that need to be hidden.
The image is rejected
Check the format and size. Free input supports JPG, PNG and WebP up to 15 MB. And very large images may be too much for the current device.
The OCR model fails to download or the run stalls
Check your connection and use Retry.
The app also has status messages such as Starting and Still working. And detection can be cancelled if you need to run it again.
The tool will not run on this device or browser
The app may not work everywhere. It suggests trying the latest Chrome or Edge. On a newer computer. Or using “Copy link to open on yuor phone”. If the current device can’t run it.
Export is blocked
You need at least one area to redact before exporting. Add a detected box or draw a manual box. Then try “Download redacted PNG” again.
Frequently asked questions
Can I redact document online for free?
Yes, for one image at a time. Free use supports JPG, PNG and WebP files up to 15 MB, English OCR, black redaction boxes and clean PNG export with no watermark. Free PNG export is downscaled to a maximum of 3 megapixels.
Does AI Document Redactor accept PDFs or Word documents?
No. It accepts document images only. JPG, PNG and WebP.
If you have a PDF or editable document, use a separate workflow to create an image copy only if taht fits your sharing need.
Will automatic detection find every sensitive item?
No. Automatic detection can miss text. Also, the UI tells users to check the preview before sharing. Use detected boxes as a starting point. Then draw manual boxes over anything missed.
Can I redact a scanned document?
Yes, if the scan is saved as a supported image file. A sharper, higher contrast scan works better for OCR. If no readable text is found, draw the redaction boxes manually.
Can I redact card numbers in a screenshot?
Yes, card number detection is included in free automatic detection. Still inspect the screenshot yourself. That’s becuase spacing, cropping or blur can make a card number harder to detect.
Are my files uploaded to a server?
The app states that your document never leaves your browser. And shows the privacy claim 100% private. Your data never leaves your device. Thnik about it.
It uses on device OCR with tesseract.js. If your device or browser can’t run it, the app suggests trying another supported setup.
Start with a copy of your document image, run detection, then slow down for the review pass. For a single JPG, PNG or WebP. The free workflow is enough for many sharing tasks. For batches, more languages, custom matching or full resolution output. Use Pro olny if those limits matter for the job in front of you.
.
Try It Free: AI Document Redactor
Upload a scan or screenshot of a document, auto-detect names, emails, IDs and card numbers via OCR, and…
Open AI Document Redactor →