Shift Technology Fraud Detection and deetech™: How They Fit

Shift Technology is one of the most established names in insurance fraud detection. Founded in Paris in 2014, it reports more than 120 customers in over 30 countries, and its AI agents cover fraud detection, claims processing, subrogation, coverage assessment and more.

deetech™ is a dedicated media-authenticity layer for insurers. It analyzes the photos, documents, audio and video that arrive with a claim and asks one question: is this evidence a genuine record of what it claims to show?

These are not competing products. Shift scores claims, entities and networks. deetech™ verifies the media. This article explains what each does, where they overlap, and how carriers can run them together.

What Shift Technology Does

Shift’s platform is built around the claim as a whole: the structured data, the text, the documents, the people and the providers connected to it.

Claims fraud scoring. Shift analyzes claims data, claimant history and provider information to flag claims that warrant investigation, so SIU teams can prioritize their caseload.
Entity and network analysis. It links claimants, providers, vehicles and other entities to surface organized activity across claims. Its Insurance Data Network extends that analysis across participating carriers.
Document fraud detection. Shift says its AI-powered document fraud detection is deployed with numerous carriers (Shift, March 2026). A 2024 case study describes it using image-similarity scoring to catch the same photo reused across claims (Shift, 2024).
Claims automation and agents. Shift Claims applies agentic AI to assessment, prioritization and automation across the claims lifecycle.
Claims-system delivery. Shift integrations for Guidewire ClaimCenter are listed on the Guidewire Marketplace, so fraud alerts can surface directly in the claim file.

In short, Shift is a claims fraud platform that checks documents and can flag a reused photo. It does not publicly document deepfake or synthetic-media detection for images, video or audio (as of September 2026). deetech™ fills that gap, and matches reused photos as well.

Where a Dedicated Media Layer Adds Depth

Generative AI has made fabricated evidence cheap to produce. Sumsub’s Identity Fraud Report found that global deepfake incidents grew tenfold from 2022 to 2023 and a further fourfold from 2023 to 2024. Deloitte’s Center for Financial Services projects that generative-AI-enabled fraud losses in the United States could reach US$40 billion by 2027, up from US$12.3 billion in 2023 (Deloitte). That forecast focuses on banking, but the same tools are available to anyone filing a claim.

A claims fraud platform treats media as one input among many. A media-authenticity layer treats each file as the subject of its own examination. deetech™ adds:

Every major evidence type in one API. Image analysis, document analysis, and uploaded audio and video analysis, so dashcam footage, recorded statements and voice notes get the same scrutiny as photos.
Document checks built for claims paperwork. Deterministic structural, arithmetic and fabrication checks, edit-history analysis, confidence-gated OCR, and pixel forensics on supported scans. These catch altered invoices, quotes and certificates. Document analysis looks for manipulation. It does not verify who a person is.
Provenance. C2PA content-credential verification reports whether a file carries valid credentials. Missing credentials are treated as unknown, not as proof of a fake.
Reused-photo matching. Every uploaded image is fingerprinted and matched against your organization’s earlier submissions, flagging exact and near-duplicate copies (resized, re-compressed or lightly edited) that turn up again in another claim. Reverse image search through configured providers then checks for photos lifted from the web.
Forensic PDF reports that set out findings, methodology and limitations for SIU investigators and retained experts.
A case workspace where investigators organize evidence and analyses for a claim.
Live channels on Enterprise plans. Live call detection and live meeting detection on Zoom, Microsoft Teams and Google Meet, for claims interviews and video inspections.

Every deetech™ output is a review input for a qualified person: it supports adjuster and SIU decisions rather than making them.

How They Work Together

In a combined stack, each system does what it was built for:

A claim is submitted with structured data and attachments.
Shift scores the claim, links entities and networks, and applies its document checks.
deetech™ analyzes the attached media through its REST API. Images often return in seconds. Video and deeper analyses run asynchronously and can take minutes, with signed webhooks notifying your system when results are ready.
Your claims or SIU system shows both signals side by side. A claim that looks routine on data but raises media concerns, or the reverse, is routed for review.
Investigators decide. They use Shift’s fraud indicators and deetech™‘s media findings and reports as inputs to their judgment.

The two signals are complementary:

Network-led fraud with genuine media. A ring using real photos across fabricated claims is a network problem. Shift’s entity analysis is built for it; media analysis will rightly find nothing wrong with the photos.
Clean data, fabricated evidence. An otherwise ordinary claim supported by an edited damage photo, a synthetic voice note or an altered repair invoice is a media problem. That is where a dedicated media layer earns its place.
Both at once. When suspicious patterns and media concerns coincide, the combined signal gives SIU a stronger basis for prioritizing the case.

Integration Notes for Shift Customers

Adding deetech™ does not require changing how Shift is deployed:

Independent operation. deetech™ analyzes media independently. Neither system needs the other’s data to function.
API and webhooks. deetech™ exposes a REST API and signed webhooks. Your team or implementation partner connects results to the claims system or SIU case view you already use.
Batch submissions. The batch API accepts up to 20 image, audio or document files per request on standard plans and up to 100 on Enterprise. Video is submitted individually.
Deployment options. Cloud by default. On-premises, air-gapped and private deployment are available on Enterprise plans for carriers with strict data-governance requirements.
Scoped access. Role-based access, scoped API keys and tenant-scoped audit logging support controlled use across claims and SIU teams.

Questions to Ask About Your Media Coverage

Whether you run Shift, another fraud platform, or both, these questions show where your media coverage stands:

Which evidence types are examined today? Photos and PDFs, or also video, audio recordings and voice notes?
How are altered documents caught? Is it by checking amounts and providers against data, or by examining the file itself for structural and edit-history anomalies?
What happens when data looks normal but the evidence is fabricated?
Can an investigator explain a media finding to a manager, a retained expert or a tribunal, using a report that sets out methodology and limitations?
Are live claims interviews covered? Synthetic voices and faces can appear in calls and video inspections, not just uploaded files.

Conclusion

Shift Technology is a proven claims fraud platform. It scores claims, maps networks, and includes document-fraud detection of its own.

deetech™ is the dedicated media-authenticity layer that sits alongside it. It examines photos, documents, audio and video in depth, verifies content credentials, produces forensic reports for qualified review, and, on Enterprise plans, extends to live calls and meetings.

Carriers don’t have to choose. Shift covers the claim and its network; deetech™ covers the evidence.

For a similar look at another claims fraud platform, see deetech™ and FRISS. For a wider market view, see the top deepfake detection tools for insurance.

To see how deetech™ helps insurers and claims teams verify the media behind every claim, visit our solutions page or request a demo.