How to remove personal information from a photo

Photos carry more personal information than most people realise. A picture of a parcel shows a full name and a home address. A holiday snap catches a stranger's face and a car plate. A photo of a form shows a date of birth and a national ID number. And even when the visible parts look harmless, the file itself often stores the exact GPS coordinates where it was taken. This guide covers how to remove all of it, both the visible data and the hidden data, before you share.

What counts as personal information in a photo

Before removing anything, it helps to know what you're looking for. In practice it falls into three layers:

• Text in the image. Names, email addresses, phone numbers, postal addresses, card and IBAN numbers, national IDs, order and account numbers, dates of birth.
• Visual identifiers. Faces, licence plates, house numbers, badges, screens in the background, signatures.
• Metadata you can't see. EXIF data embedded in the file: GPS coordinates, the date and time, the device model, sometimes the owner's name and a thumbnail of the original image.

Removing only the first layer is the most common mistake. A photo with a perfectly blacked-out address can still tell anyone exactly which house it was taken in, down to a few metres.

The fastest way: remove it automatically

Finding every piece of text in a photo by eye is slow and easy to get wrong. An automatic tool reads the image for you and boxes what it finds. With Privshot:

• 1. Drop the photo in. PNG, JPEG or WebP.
• 2. Let it detect. Names, emails, phone numbers, addresses, cards, national IDs and more are found and covered automatically.
• 3. Review. Add a box over anything it missed, such as a face, a plate or a background screen, and resize or remove any box you don't want. Nothing is exported until you say so.
• 4. Download. You get a flattened PNG: the covered pixels are gone, and the export carries no EXIF from the original.

The photo is processed in memory and discarded straight away, so nothing is stored or used to train a model.

Doing it manually

If you'd rather not use a tool, you can remove personal information by hand, as long as you flatten the result:

• iPhone / iPad: Photos, then Edit, then Markup, then draw a solid, opaque rectangle over each sensitive area. Avoid the highlighter, which is translucent.
• Android: Google Photos, then Edit, then Markup, using the pen at full opacity.
• Windows: Paint or Snipping Tool, a filled black rectangle, then Save as a new PNG.
• macOS: Preview, then Markup, then a filled rectangle, then File → Export to a new PNG.

The same rules apply as for screenshots, and we go into them in detail in how to redact a screenshot.

Don't forget the metadata

Covering pixels does nothing to the EXIF block. To strip it:

• iPhone: in Photos, tap Share, then Options, then turn off Location before sending.
• Android: in Google Photos, Share, then Remove location.
• Windows: right-click the file, then Properties, then Details, then Remove Properties and Personal Information.
• Anywhere: re-export or screenshot the image. A freshly written PNG inherits none of the original EXIF.

The mistakes that leave the data recoverable

• Blurring or pixelating. Both are reversible far more often than people expect, especially for short structured data like a card or phone number. See why blurring is not safe redaction. Use a solid fill.
• Emoji and stickers. A sticker dragged over a face is still a movable layer in some editors, and it rarely covers the full area.
• Cropping instead of covering. Cropping is safe if you export a new file. Some apps store the crop non-destructively, so the full frame is still in there.
• Partial coverage. The tops and tails of letters are often enough to reconstruct a word. Cover the whole field, with margin.
• Reflections and background screens. A monitor, a mirror or a pair of glasses can carry readable text.
• Sending the original by mistake. Save the clean copy under a different name.

Removing personal information at scale

If photos arrive continuously, whether from user uploads, support attachments, KYC documents or moderation queues, manual work stops being viable. Privshot's REST API detects and removes personal information in images programmatically: you send one request and get a clean image back. There's a free tier to try it, and a Python walkthrough to get going in a few lines.

In short

To remove personal information from a photo properly: cover every piece of text and every visual identifier with a solid fill, never a blur; export a flattened file so the covered pixels are truly gone; and strip the metadata so the file doesn't carry the location home. An automatic tool does all three in one pass, so try it here.