AI Transparency¶
Statement¶
ClaimGuard's product mission is to help insurers detect AI-generated and manipulated media in claims. The detection function is implemented on-VM with internal analysis services that perform pattern detection, forensic analysis, and provenance verification on uploaded images.
There is no external generative-AI integration in current production scope. ClaimGuard does not, in the current production configuration, send customer claim data to a third-party generative-AI inference service. Any future change that introduces such an integration will be disclosed via the trust portal change process and added to the Subprocessors page before going live in production, with at least thirty days' notice to affected customers.
Implementation¶
Where AI / ML is used in the product¶
The product uses detection-side machine-learning: image-analysis, manipulation-detection, and content-provenance verification services that run locally on the production VM. These services are pattern detectors and classifiers — not generative AI — and they do not call out to external generative-AI services.
What is sent externally¶
For the current production configuration: only the GCP control plane (Secret Manager, Cloud Logging, Cloud Storage), per the Subprocessors page. No customer claim data, no narratives, and no evidence is sent to any external generative-AI service.
Customer-facing transparency¶
- No automated decision-making. Every claim assessment is presented to a human reviewer; the reviewer is the decision-maker.
- Advisory output labelling. Detection-service output is labelled in the application UI as advisory and not as a determination.
- Customer notification of any new external generative-AI integration. Per the Vendor management process, any new production-data-handling vendor — including any future generative-AI inference vendor — is reviewed, added to Subprocessors, and customers are given the contractually-required notice before the integration goes live.
Status¶
implemented — verified 2026-05-10. No external generative-AI integration in current production scope. The product detection function uses internal on-VM ML services only. Any future change is gated by the trust portal change process and the Subprocessors update commitment.
Roadmap¶
- Customer-facing notice template for any future external generative-AI integration — to be drafted before any such integration is introduced.
- AI risk inventory (AI risk management) to be populated with the relevant risk-mitigations whenever an external generative-AI integration is introduced.