About
Built for the people who check the evidence
VeriSight helps compliance and audit teams take a closer look at the images they're asked to trust: receipts, invoices, badges and meeting photos submitted as evidence.
Why it exists
A photo is easy to fake now
Expense claims and meeting records increasingly rest on a photo. Making a convincing receipt or changing a total used to take skill and time. Today an image generator, or a few minutes in an editing app, is enough, and the result is hard to tell apart by eye.
Most reviewers aren't image forensics specialists, and they shouldn't need to be. We built VeriSight to give them a fast, structured first look: three independent checks in a few seconds and a report they can file. It doesn't replace their judgment. It shows them where to look and what to ask for next.
What it does
Three checks, one report
- Check 1
The AI check
Scores how likely the image is to be AI-generated or AI-edited, and places the score in one of four bands.
- Check 2
The tamper check
Outlines regions that look edited with image software, such as a typed-over total or a pasted field, each with a confidence.
- Check 3
The visual review
Four views computed from the image that make mismatched ink and resaved regions easier to spot, with notes on how to read each one.
How we work
Three principles
- 01
Evidence, not verdicts
Every result is a starting point for a reviewer. We show scores, regions and views with plain notes on what they can and can't tell you, and we don't merge them into a single pass or fail.
- 02
Nothing is kept
No accounts, no database, no file storage. An image is held in memory while it's analyzed and discarded afterwards, and the report lives only in your browser.
- 03
Honest about limits
Detection can miss and it can misfire. We list where it falls short on the homepage, so nobody mistakes a score for proof.
Have an image to check?
It takes a few seconds, and nothing is stored.