TrustGraph

A privacy-preserving trust intelligence network that helps institutions spot repeat fraud across organizations, without pooling anyone's customer data.

Fraudsters rarely attack just one company. TrustGraph lets participating institutions learn that a face or identity has shown up as a problem elsewhere, while each institution's customer records stay where they are.

A network-based fraud detection capability from Incode in which participating institutions query de-identified signals across the network and receive an explainable result, while raw customer data stays inside each institution's own environment. The network runs on secure multi-party computation, a cryptographic method that lets several parties compute a joint answer without any of them revealing its inputs, and it has no central database of consumer records.

A TrustGraph check typically works like this:

1. The institution verifies the person's identity as usual.
2. It queries TrustGraph privately, with identifiers de-identified before they leave its environment.
3. The network compares the signals across participating institutions using secure multi-party computation.
4. The institution receives an explainable signal and decides what to do with it.

Several design choices define how it works:

• No central data store, so there is no pooled database of customer records for an attacker to target.
• Four privacy guarantees, which hold on every query and are described on the TrustGraph privacy guarantees page.
• Institution-controlled decisions, since the signal informs the institution's own rules and review, and TrustGraph does not make the decision.
• Shadow mode by default, so customers can see the signals before they act on them.

TrustGraph runs on Incode's privacy-preserving cryptography, strengthened by Identiq's patented technology.

TrustGraph sits next to terms that are easy to mix up:

• Fraud consortium is a shared-data model in which records move to a central place, while TrustGraph keeps raw data in place.
• Identity graph links accounts inside one organization's data, while TrustGraph looks across institutions.
• Identity verification confirms who a person is at the start, and TrustGraph adds a signal about whether that identity has been a problem elsewhere.