Redacting a Screenshot to Protect Whoever Leaked It to You

Every other page on this site assumes you're protecting yourself, or someone who appears in your own photo. This page is about a different person entirely: whoever handed you the screenshot in the first place. Someone sent you an internal message, a leaked document, a private chat, or an email they weren't supposed to forward, and now you want to share it, post it, or publish it — but if it's traced back to where it came from, that person could lose a job, face legal exposure, or worse. Covering the sensitive content in the frame is still worth doing. It just isn't the thing that protects them.

That's a structurally different problem from redacting your own photo before you post it. There, the person in the image and the person deciding what to cover are the same, and the threat model is usually casual: a stranger scrolling past, not someone actively trying to reverse-engineer your identity. Here, the two people are different, the source has already taken a real risk to get you this material, and whoever it exposes — an employer, an agency, a company — may be specifically motivated to work backward from the published image to figure out who leaked it. That's an adversarial threat model this site hasn't addressed anywhere else.

And the identifying details in play usually aren't inside the frame you'd think to cover. They're in the filename, in metadata nobody looked at, in artifacts the original document already carried before it reached you, or in the simple fact that out of a handful of people with access, only one copy looked exactly like the one that leaked.

Drop your image here

Or click to browse · Paste with Ctrl+V also works

PNG · JPG · WebP · GIF

How It Works

Open

Drop your image in or paste from clipboard.

Pick Mode

Black Box, Blur, or Pixelate.

Select Areas

Rectangle, oval, or freehand lasso — then hide what you selected.

Download

Hit Download PNG. Done.

SquooshNeed to shrink your image after editing? Squoosh is a free browser-based image compressor with no upload required.

Visit Squoosh →

Guide

One sentence carries this page: covering what's sensitive in the frame protects the subject of the image, not the person who gave it to you, and those are two different jobs with two different failure modes.

Why protecting a source is a different redaction problem

Almost everywhere else on this site, the person deciding what to cover and the person the image is about are the same, and the realistic threat is casual: a stranger scrolling past a screenshot, not someone methodically trying to reverse your identity. A leak flips both of those. Someone else took a real risk to get you this material, the image is about something they exposed rather than about them personally, and whoever it embarrasses or implicates — an employer, an agency, a company's legal team — has a direct incentive to work backward from what got published to figure out who was responsible. That's an adversarial, motivated search, not an accidental glance, and a region-only cover was never built to defeat one.

What actually gives up who gave it to you

The things that identify a source rarely sit inside the area you'd think to black out. A file's original name, kept intact by most sharing tools unless you specifically rename it, can embed a device name, an internal project code, or a timestamp from the source's own system. Metadata most image viewers never show by default can carry similar information. Documents that passed through a color printer before they were photographed or scanned can carry tracking dots — a near-invisible yellow pattern many printers add to every page, encoding the printer's serial number and the time of printing, that survives a photo of the page just fine. According to Wikipedia's account of the Reality Winner case, which cites reporting from Vice News, journalists and security experts have pointed to exactly this: The Intercept published scans of a leaked NSA document that still showed its original tracking dots, and that, combined with an internal audit narrowing access to six people, is widely cited as a factor in how Winner was identified and later convicted in 2017. None of that would have been touched by covering a name or a paragraph in the image itself.

When the copy itself is the fingerprint

There's a subtler version of the same problem. If several people had access to the source material, and each saw a copy that was even slightly different — a different crop, a different page order, a different resolution, a different exact wording — then the specific copy that leaks can point straight back to whoever had that copy, independent of anything inside the frame. Deliberately giving each recipient a uniquely distinguishable version specifically so a later leak can be traced is a real, long-documented practice sometimes called a canary trap or barium meal test; intelligence services have used versions of it for decades, well before the term reached fiction. You usually can't know whether the copy you were handed was individually marked this way. What follows from that is simple: redacting the part you can see says nothing about a distinguishing property you can't.

What this editor covers, and what it doesn't

Read directly from what this tool actually does, not what it's convenient to assume it does. It operates on a region you select, with three modes — Black Box, Blur, Pixelate — across rectangle, oval, and lasso shapes, and nothing else. It has no code that reads or edits file metadata, no code that touches a filename beyond what happens automatically on export, and no way to detect or strip printer tracking dots, since those live in the pixels themselves rather than in any field a region tool can see. One thing it does do that happens to help here: every download is saved under a generated name — hideshot- followed by the current timestamp — so whatever the original file was called, that name doesn't travel with your exported copy. That's a side effect of how the download works, not a privacy feature built for this purpose, and it only touches the filename; everything else in this section still applies.

A practical ordering before you publish

Ask what you can about how the material reached you before you decide it's safe: was it printed at any point, and if so, do you or the source control which printer; did different people see different versions; does the file you have still carry its original name or metadata. Strip metadata and rename the file yourself rather than relying on any single tool to do it as a byproduct. Use Black Box, not Blur or Pixelate, for anything genuinely sensitive in the frame, since this is an adversarial read, not a casual one, and a cover that only degrades detail invites someone determined to try reversing it. Decide what you're willing to say about how you obtained the material before anyone asks, since an answer worked out in the moment is more likely to contradict itself than one you prepared. And don't assume anyone upstream of you — the source, a previous recipient, a printer in the chain — already handled any of this; check it yourself.

Common mistakes and misconceptions

"I blacked out the name, so the image doesn't point back to anyone." It doesn't point back to the subject anymore. The filename, the metadata, and anything baked into the original document are all untouched by a region you drew over the image, and any one of them can point back to the source instead.

"Printer tracking dots are old news; outlets fixed this after 2017." Some outlets got more careful. That tells you nothing about the specific printer, source, or previous recipient the material you have passed through before it reached you. Treat it as unverified until you've checked, not as solved because one well-known case made headlines.

"If I crop out the tracking dots, I'm fine." Only if you're certain the dots were the only distinguishing property. A canary-trapped copy can be marked in ways that have nothing to do with a visible dot pattern — a unique crop or resolution can itself be the marker, which means cropping doesn't necessarily remove the fingerprint; it can just move where it sits.

"This only matters for journalists handling classified leaks." The same structure shows up any time the person who gave you something is not the person it's about: a coworker forwarding an internal email, a tenant sharing a landlord's message, an employee passing a memo to a regulator. Whenever those two people differ, you're protecting both, and a region-only redaction only ever reaches one of them.

A Person This Editor Can't See

Covering a name or a face in this tab protects the subject of the image. If someone handed you that image at real risk to themselves, the source, it does nothing for them, because a region tool has no way to touch a filename, hidden metadata, or artifacts like printer tracking dots that were already part of the file before you opened it. Those are exactly the kind of detail that has reportedly contributed to identifying a source before — according to Wikipedia's account of the Reality Winner case, which cites Vice News reporting, journalists and security experts have pointed to The Intercept's publication of tracking-dot-bearing scans, alongside a narrow internal access list, as a factor in how she was identified in 2017.

There's a subtler version of the same risk: if different people were given subtly different copies of the same material specifically so a leak could be traced back to whoever had it, redacting the visible content does nothing about that either. This technique, sometimes called a canary trap, is real and predates the term by decades.

The practical response is to strip metadata and rename the file yourself, use Black Box rather than Blur or Pixelate for anything sensitive since this is an adversarial read rather than a casual one, and check what you can about how the material reached you — was it printed, did other people see different versions — before deciding a cover in this tab is enough.

Frequently asked questions

Does covering the sensitive part of the image also hide how it reached me?

No, and that gap is the whole point of this page. A black box, blur, or pixelate region only changes pixels inside the area you select. It does nothing to the exported filename, nothing to any metadata the original file carried, and nothing to artifacts that were baked into the image before you ever opened it, like the near-invisible tracking dots many color printers add to every page. Protecting what's visible in the frame and protecting who handed you the frame are two separate jobs, and a tool built for the first one has no way to do the second.

Didn't newsrooms already learn this lesson after the Reality Winner case?

Some did, but you can't assume everyone handling a leak before it reaches you has. In 2017, Reality Winner was convicted of leaking a classified NSA document to The Intercept. According to Wikipedia's account, which cites reporting from Vice News, journalists and security experts have pointed to The Intercept's decision to publish scans of the document that still showed its original printer tracking dots, along with the fact that only six people had accessed the document internally, as factors in how she was identified. Awareness of the risk at one outlet in 2017 doesn't mean a coworker, a source, or a printer somewhere in the chain today has stripped anything. You generally have to assume nobody upstream of you has handled this and check yourself.

Could publishing my own redacted crop expose which copy I was given?

It can, if whoever gave you the material deliberately varied what different recipients received. Giving each recipient of a sensitive document a slightly different version, so that a later leak can be traced back to the specific copy that surfaced, is a real and long-documented practice sometimes called a canary trap or barium meal test, used by intelligence services long before the term was popularized by fiction. It isn't limited to swapped words in text. A unique crop, a distinctive framing, or even a specific resolution can serve the same tracing purpose in an image. If you can't rule out that your copy was individually distinguishable, redacting the sensitive region doesn't address that risk at all.

Is this only something journalists need to think about?

No. The same problem applies any time someone hands you something they weren't supposed to share and could face real consequences, professional or legal, if it's traced back to them: a coworker forwarding an internal message, a tenant passing along a landlord's email, an employee sharing an internal memo with a reporter or a regulator. Whenever the person you got it from is not the same person the image is about, you're protecting two different parties with two different threat models at once, and a region-only redaction only ever handles one of them.