Live demo
Is this document what it claims to be?
Run four of ScopeFlow's forensic signal families on an image or PDF. The analysis happens in your browser: the file never leaves your device.
No file at hand?
Generate a synthetic specimen in your browser and run the check on it.
Runs entirely on your device. Nothing is uploaded.
What the checks mean
Four signal families, explained
Error-level analysis
The image is re-saved as a JPEG and compared with itself. Areas that were pasted in or retyped often react differently from the rest of the page, because they have a different compression history.
Metadata
Camera make and model, capture date and the software that last saved the file. An official document last saved by a photo editor is worth a second look.
Provenance & AI markers
Content Credentials (C2PA), IPTC 'AI-generated' source types, and the parameters some generators write into the file.
PDF structure
Producer software, incremental saves after the first version, modification dates, duplicated font subsets and digital signatures.
Honest limits
What a single check can't tell you
- Metadata is easy to strip or forge, so its absence proves nothing.
- Error-level analysis is noisy on screenshots, resized images and heavily textured photos.
- A cleanly regenerated document can pass every pixel check. That is why the full platform adds issuer verification, consistency checks and cross-case linking.
- No single check is a verdict. Treat results as pointers to what to inspect in the original.
Now imagine this on every file, linked across your whole caseload
ScopeFlow runs dozens of signals on every document at intake and connects cases that share templates, sponsors and addresses.