ClaimWard turns any phone camera into a source of certified, court-grade evidence — for any claim, return, inspection or condition report a photo has to prove. Sealed at the moment of capture, delivered on your own domain. No app to install. EU-sovereign by design.
AI-generated damage, doctored number plates and recycled photos are flowing into claims faster than review teams — or detection alone — can keep up. The evidence layer needs to change at the source.
+300%
rise in manipulated-image claims at one major European insurer over a single year
Allianz UK · The Guardian
~10%
of property & casualty claims are fraudulent — with images now a primary vector
FBI / industry estimates
6.5%
of all fraud attempts globally now involve deepfakes — up 2,137% in three years
Sumsub, 2025
2 Aug
EU AI Act Article 50 transparency duties apply in 2026 — synthetic media must be marked
EU AI Act
How it works
A code, a camera, a sealed bundle.
The customer never installs anything. They open a link on your own domain and follow a guided, three-minute capture. Every step is also a security control.
STEP 01
Issue
You send a single-use code — by email, dashboard or API. Zero IT to start.
STEP 02
Open
The customer opens claim.yourbrand.com in any browser. No app.
STEP 03
Capture
A guided photo & video session — with no way to upload from the gallery.
STEP 04
Seal
Each frame is hashed, C2PA-signed, timestamped and eIDAS-sealed into a bundle.
STEP 05
Deliver
You receive a trust score, the checks that ran, and a public verification link.
Screen-replay check · walkaround
MONITOR = FLAT PLANE
device motion (IMU)optical flow
Designed to be hard to fool
It knows a screen from a scene.
The fraudster's favourite trick is photographing a fake image off a high-resolution monitor. ClaimWard's strongest defence is nearly free in a browser: during a short guided walkaround, the phone's motion sensors describe its real path while the video's optical flow describes the scene. A real object has depth and parallax — a monitor is a flat plane, and the two signals can't stay consistent.
Randomised shot scripts — the angles are chosen at capture time, so nothing can be pre-generated.
A live, handwritten nonce — a fresh code shown on screen must appear, handwritten, in the next shot within 90 seconds.
Forensic layers — macro texture, torch response, moiré and GenAI-artifact analysis run server-side.
Honest by design. A browser can't do hardware attestation, so ClaimWard sells calibrated assurance levels — never "unhackable." With insurance buyers, that candour is the point.
1
Hashed at captureSHA-256 per frame, in the browser — before anything is uploaded
2
Chained & C2PA-signeda Merkle root over every frame, sensor stream and check result
3
Qualified timestampRFC 3161 from an EU-listed trust service — legally recognised time
4
eIDAS qualified sealoptionally under your own certificate — EU-wide legal presumption
5
WORM storage in the EUwrite-once storage in the EU — or fully on-premises
6
Published to a transparency logan append-only, publicly auditable Merkle log with witnessed checkpoints — anyone can prove the evidence existed and was never altered
Provenance, not promises
Sealed the instant it's captured.
Every capture becomes a tamper-evident bundle with a chain of custody that stands up outside your walls. The differentiator is where it's sealed: qualified timestamps and qualified electronic seals under eIDAS carry an EU-wide legal presumption of integrity and origin — something no overseas tamper-score can assert.
Beyond blockchain — deliberately. Evidence hashes are anchored in an append-only, tamper-evident transparency log: the architecture that has secured the web's certificates for over a decade. Unlike a blockchain, there is no consensus overhead and nothing private about it — the log is publicly auditable, and its checkpoints are countersigned by independent witnesses, so the proof doesn't depend on trusting any single operator. Including us.
Processed and stored in the EU — with a full on-premises option for sovereignty mandates.
GDPR-minimal — location is opt-in, device identifiers are hashed, retention is yours to set.
Verifiable beyond trust in us — bundle hashes are published to an append-only transparency log (the Certificate-Transparency pattern), so integrity can be proven against independently witnessed log roots, not our word.
Why ClaimWard
The European seat, taken.
The category was pioneered in the US. ClaimWard is built for how European insurers actually have to operate.
EU-sovereign by default
Data residency, eIDAS sealing and on-prem deployment — not a bolt-on. The first service to combine them for claims.
On your own domain
Customers see claim.yourbrand.com, your logo and — at enterprise tier — your own seal. Not ours.
Live the same day
The email tier needs zero integration. Send a code, get sealed evidence back — before IT is even involved.
Built for claims
Plate, VIN and serial numbers are read and bound to the claim; recycled images are caught across every tenant.
Transparency, not trust
Every sealed bundle is anchored in a public, append-only transparency log — the pattern that secures the web's certificates, without a blockchain's overhead or opacity. Court-grade proof that doesn't depend on believing us.
Governed by the suite
ClaimWard runs its own AI checks through Gateward and under RiverAct governance — we run on our own stack.
Who it's for
Insurance is the wedge — not the ceiling.
Anywhere a photo or video proves a claim, the same AI fraud applies — and the same sealed, court-grade capture defends it. ClaimWard drops into any workflow that has to trust what a camera saw.
ClaimWard is onboarding a small group of European insurers, TPAs and loss adjusters — plus early partners in warranty, rentals, marketplaces and lending — ahead of the EU AI Act's August deadline. Request access and we'll set up a pilot on your own domain.
