Screenshots are not evidence: teaching provenance

In children's digital social life, the screenshot is the unit of proof: look what she said (a chat capture), look what I got (a game score), look what happened (a headline). The entire evidentiary weight rests on an assumption nobody examines: that a screenshot shows what a screen really showed. It doesn't have to — and the fabrication requires no tools beyond the browser itself.

The technical fact, stated plainly

A screenshot is a picture of pixels, and every pixel on a screen is negotiable before the picture is taken: any page's text can be edited in place with the browser's built-in developer tools (two clicks, no software), chat apps render whatever a template mimics, and generator sites mass-produce fake conversations for free — before AI image tools made even the picture of the pixels synthesizable. None of this is exotic knowledge; it is exactly the demonstration a classroom can do live. A screenshot, evidentially, is testimony, not evidence: it's the screenshotter making a claim, and it inherits their credibility — who says this? — not the credibility of the screen it depicts.

What real provenance looks like (and doesn't cover)

Provenance technology exists: the C2PA / Content Credentials standard cryptographically binds capture-and-edit history to media files, and camera makers and AI generators are adopting it. Worth knowing about, honestly limited: adoption is partial, credentials strip when content is re-screenshotted or re-encoded (absence of credentials proves nothing), and chat screenshots — the case children actually litigate — sit entirely outside it. For the playground dispute, the working provenance checks are social, not cryptographic: does the original exist on a device, in the app, in context? Does a second witness's copy agree? The second-source reflex, pointed at receipts.

The double lesson

Screenshot literacy runs in both directions, and the second is the safety-critical one:

1. Don't trust screenshots — they're claims wearing evidence's clothes.
2. Don't trust them to not exist — anything you send can be screenshotted, and "disappearing" messages disappear from the app, not from the world (permanence). The same fact that makes screenshots weak evidence makes them strong ammunition — a captured cruel message is forever, fake or real, which feeds both group-chat dynamics and cyberbullying patterns.

The two directions share one root — pixels are copies, copies are editable, copies are keepable — and teaching them together is more coherent than either alone.

What the evidence doesn't say

• It doesn't quantify how often children are fooled by fabricated screenshots — the prevalence is undocumented; the capability and the social role are what's established.
• It doesn't make screenshots worthless — most are honest, and they remain useful testimony; the lesson is calibration ("screenshot + trustworthy sender + original checkable" is fine), not blanket dismissal.
• It doesn't let provenance tech carry the weight yet — teach the social checks; mention the cryptographic future.

In the classroom

1. Do the live edit (teacher's machine, fictional page): change a headline with devtools, screenshot it — thirty seconds that permanently recalibrates a class.
2. Install the response norm: a dramatic screenshot earns the question "can I see the original?" — asked without accusation, as routine as citing sources.
3. Teach the both-directions rule together: can't trust them, can't outrun them — one lesson, two warnings.
4. Route disputes to adults: a screenshot dispute between children (fake receipt, leaked capture) is a trusted-adult moment, not a peer trial.

How Wiz Kids applies this

Judgment scenarios in the safety strand present the screenshot cases both ways: the too-dramatic "proof" that warrants asking for the original, and the "it disappears after you send it" claim that doesn't survive the screenshot fact. The simulated apps make the underlying point mechanically — children see that pages are just rendered text a tool can change — and the permanence lessons carry the second direction.

References

• C2PA. Content Credentials specification (c2pa.org) — the provenance standard and its scope.
• Wineburg, S., & McGrew, S. (2019). Lateral reading. Teachers College Record, 121(11) — verification by leaving the artifact, the transferable method.
• Lewandowsky, S., et al. (2012). Misinformation and its correction. PSPI, 13(3) — why a vivid fake outlives its debunk, the stakes behind the skepticism.

© Glu IO Pty. Ltd. — Wiz Kids (wiz.kids). Link freely; republication requires permission — see terms. Found an error in our reading of the research? We correct fast: tell any teacher piloting Wiz Kids.