VARNIUMAI risk defense / everyday bearings

DEEPFAKE / RESHARE TRIAGE

Before You Reshare an Image: A Three-Source Triage

A single image can be real, recycled from another year, or generated — and the caption decides how it travels. Resharing is an act of publication, so it deserves the same three minutes a reporter spends.

Start here

The free tool below runs entirely in this browser — no account, nothing stored or transmitted. Redact anything you would not want kept before you type it.

Open the companion tool →

Why a three-source rule beats a gut check

Images get recycled constantly: a real photograph from a different year or country, a still from a film, a generated picture with a plausible caption. Each of those fails a different one of the three checks, which is why one check is not enough.

The failure mode is not being fooled — it is passing the image on with certainty you do not have. Naming what you could not verify keeps the reader honest and still lets useful, verified material travel.

The protocol

  1. Find where the image first appears. Search for the image itself and read the oldest instance you can find, noting its date and location.
  2. Check the outlet or agency the caption names, on its own site. If an organization took the photo, it usually publishes it — and if it did not, that absence is informative.
  3. Check whether the content credentials or file information claim generation or editing, and treat their absence as unknown rather than a clean bill of health.
  4. Look for one detail in the image that must match the caption: signage, weather, uniforms, vehicle plates, or season. Mismatches are the most common tell.
  5. Decide the outcome: share, share with a caveat, or don't share. All three are legitimate; only the silent fourth — share with confidence you don't have — is not.

What the tool does

Run a three-source triage and get one of three honest outcomes for the image in front of you, with the checks you ran written down. It stores nothing and sends nothing.

Limit first: This is a triage aid, not an image-forensics tool and not a fact-checking service. It cannot prove an image is real or generated; it can only keep you honest about what you checked.

Open the companion tool →

One human next step

If a manipulated image of you or your business is circulating, save the original and the posts, report to the platform, and note the account details before anything is deleted. For commercial harm, a lawyer who handles defamation can advise on next steps.

Research log and safety checks

AI-assist path: not used for a verdict. Image authenticity is presented only as sourced or unsourced, per §1.2 UIC-10 — the artifact's value is the three-source discipline, not a probability score. §6.10: not a frontier-pace topic.

§0.5 protection result: category is deepfake-media-verification; crisis-adjacent: no; fear/urgency/scarcity toolkit used: no. The protective function is free and the paid feature is convenience only.

Related dispatches

Pairs with the same protection bar: a free complete reading, an optional convenience unlock, and no fear-framing.