Screenshots get shared with less caution than photographs. A screenshot of a receipt, a chat, a map, a settings page — it feels like a picture of a screen, not a picture of a place. That instinct is half right. The metadata is thinner. The content is usually worse.
It helps to separate the two questions completely: what is stored in the file format, and what is drawn into the image. They have different fixes, and only one of them is a metadata remover.
A screenshot is a rendering
When you capture a screen, the operating system reads its framebuffer and encodes it as PNG. No camera is involved, so the fields a camera would write — exposure, focal length, GPS, serial numbers — never exist in the first place.
What the capture tool writes instead is producer information: which application saved the file, which OS version was running, and when. Some tools add a title or a caption. None of that is location data, and all of it is still information about you.
What is actually in the bytes
PNG stores metadata in chunks, each with a four-letter type. The text chunks are the ones that matter for screenshots, because they are free-form key/value pairs that any tool can write.
| Chunk | What it stores | What it can reveal |
|---|---|---|
tEXt |
Uncompressed key/value text | Software, Author, Comment, Title |
zTXt |
The same text, deflated | The same fields, hidden from a plain string search |
iTXt |
UTF-8 text, optionally compressed | Captions, titles, and sometimes an embedded XML metadata packet |
eXIf |
A full EXIF block | Device data, if the tool that wrote the file chose to include it |
tIME |
Last modification time | When the file was produced or edited |
iCCP |
Colour profile | Display pipeline, not identity — keep it |
Which of these appear depends entirely on the capture tool and the platform. A screenshot from one device can be almost bare, while a screenshot that passed through a chat app, an editor or a documentation tool can carry a second round of producer fields.
Software, device and app names
The Software field is the common one. It typically names the capture or editing
application, and on some platforms it includes an OS name or version. That tells a reader
which ecosystem you are on and roughly how current it is — useful for targeted phishing, and
useful for anyone building a profile of your devices.
Some tools add more. An app name, a document title, a user name from a licensing field. If you captured the screen from inside a work application, the file may remember the application's name even after you crop the interface out of the frame.
What is actually in the pixels
Here is the part that matters more. A screenshot is a complete record of a rectangle of your screen at one moment. Everything visible in it is now data: the sender name in a chat header, the email address in a sidebar, the URL in an address bar, the file path in a window title, the notification that arrived mid-capture.
Status bars add a quiet layer. Time, battery, carrier or network name, a VPN icon, the number of unread messages. Individually mundane; together they narrow down who you are, where you are, and when you were there.
Cropping removes what is outside the crop, not what is behind it. A blur or pixelation applied in an image editor is baked into the pixels only if you export the flattened result — and a weak blur can sometimes be reversed. A black box drawn over a line of text is reliable; a translucent highlight is not.
The map problem and the document problem
A screenshot of a map is a location disclosure without a single GPS tag. The pin, the street names, the labelled neighbourhood, the route line — all of it is readable by a person and by OCR. Metadata removal does nothing here, because the coordinates were never in the metadata.
A screenshot of a document is the same shape of problem. A bank statement, an invoice, a ticket, a medical report: the sensitive content is the content. If the document was on screen, it is in the PNG. If you need to share it, redact before capture, not after.
This is why "is this screenshot safe?" has no metadata-only answer. The file can be perfectly clean and still be the most revealing image you own.
Cleaning a screenshot
Strip the text chunks and any eXIf block, keep the ICC profile, and check the
result. That removes producer fields without touching pixels, so text stays sharp and colours
stay correct.
Then do the part software cannot do: read the image as a stranger would. Check the corners, the status bar, the window chrome, the reflections in any photo shown on screen. If something should not be there, recapture with it hidden rather than trusting a crop.
One last note: screenshots are often used as a "safe" way to share a photo from another app. They are safe for the original's EXIF and not safe for the original's content. Use them for convenience, not as a sanitising step.
See what your screenshots are carrying
Drop a PNG in. MetaPurge lists the text chunks and any embedded EXIF, then rewrites the file without touching a pixel.
Open the metadata removerWhere to go next
- Photo metadata privacy risks — what producer fields and timestamps add up to.
- What is EXIF metadata? — the camera side, and where PNG keeps its own version.
- Facebook photo metadata — why a screenshot is not a clean copy of a post.