Veriframe verifies still images the same way it verifies video: against the indexed original, not against a guess. Publish a photo once, and anyone holding a copy — a repost, a screenshot, a crop, an edit — can check whether it still matches what you actually released.
Why it matters
Index a photo at release time and every copy in circulation — reposts, crops, screenshots, edits — can be checked against it by anyone, without an account. The verdict comes with a permalink you can point to.
Social platforms recompress, resize, and convert everything they touch. VERA names that processing for what it is and keeps the copy anchored to your original — no false alarms on harmless copies.
An added logo, a caption bar, a colour grade, a mirror flip — the verdict states what changed and the evidence it rests on, so a correction or fact-check can quote it directly.
How VERA works for images
Your communications team publishes the original photo to the Veriframe registry — one upload, at release time. The registry stores its forensic fingerprints, not just the file.
A journalist, fact-checker, or citizen drops the copy they received into the public verify page. No account needed.
The verdict is grounded in deterministic image fingerprints compared against the indexed original — the same evidence ladder our video verification uses, applied to a single frame. What changed is named: recompression, resizing, an added logo or caption, a colour edit, a mirror flip.
What the verdict tells you
Every verdict states its evidence. These are the outcomes the image pipeline distinguishes today:
Platform recompression, format conversion, and resizing keep the image anchored to its original — reported as re-encoded, not as tampering.
A localized edit — an inserted graphic, a caption bar, a stamped logo — lights up a specific region of the image grid while the rest still matches the original.
Saturation and colour-grade changes, and the classic mirror-flip repost, are detected and named against the indexed original.
Moderate crops stay anchored to the original and are reported as such. Heavy reframing can break the anchor — in that case VERA says "not found" rather than guessing.
Use cases
Portraits, working visits, press photos: index them as they are published. When an edited version starts circulating, the press office points to a verdict instead of issuing a denial.
A photo arrives via social media minutes before the deadline. Drop it into the public verify page: if the source indexed the original, you know within moments whether the copy was altered — and what changed.
Every verification gets a permalink. A fact-check, a community note, or a reply can link straight to the evidence instead of arguing about it.
Honest limits
Veriframe proves whether an image matches an indexed original. It does not — and does not pretend to — look at an arbitrary image and rule whether it was AI-generated. That claim cannot be made reliably, so we never make it.
If the original was never indexed, VERA answers "not found" — honestly. The registry model is the point: proof, not probability.
Video originals are indexed at one frame per second. A still taken exactly on a sampled frame anchors to the video; one taken between samples returns "not found" with guidance, never a guess.
The public verify page accepts JPEG, PNG and WebP alongside video. Drop in a copy and read the evidence yourself.