Introducing TrustGraph, the first trust network built for the agentic era

Today, Incode launched TrustGraph, the privacy-preserving trust network that enables institutions to detect repeat fraud and recognize legitimate customers with signals from across the network, without exposing customer data. In Incode's early access program, institutions using TrustGraph detected up to 53% more fraud than with their existing models alone.

With TrustGraph, raw customer data stays with the institution that holds it and is not visible to other participants or to Incode. This lets institutions identify repeat fraud without exposing customer data to competitors.

TrustGraph launches as fraudsters increasingly run the same playbook against hundreds of companies at once. They conduct an attack on Company A, learn from it, and execute the same pattern on Company B. Now, when both institutions participate in TrustGraph, Company B can act on a signal about that attempt in real time.

Modern fraud scales across institutions at machine speed

Fraudsters can now use AI agents to build synthetic identities, forge documents, and retry after every rejection, targeting hundreds of companies at once. Nine out of 10 fraud professionals agree that AI-driven fraud is escalating, and Incode projects that AI will drive more than 60% of fraud attempts by the end of 2026.

AI has democratized fraud

Generative AI has made fraud cheaper and easier to commit than ever before, and criminals can now buy deepfake images, synthetic identities, and cloned voices for as little as $5. Convincing AI-generated deepfakes easily imitate a real person's face or voice, and AI-based injection attacks go one step further by feeding fake camera footage straight into a verification session.

Fraudsters can produce all of these at scale. An attack that once took a fraud crew one week takes less than an hour.

How agentic AI shortens attack timelines

AI agents run those capabilities at machine speed. An agent can create a synthetic identity, forge a document, and attempt verification at hundreds of institutions in parallel, around the clock. When one institution rejects an attempt, the agent learns from the rejection, adapts, and retries. That gives attackers a speed advantage: agentic fraud adapts in an afternoon, while a fraud model retrains in a quarter.

Fraud used to scale with people, and then it scaled with software. Now, it runs itself.

Why siloed defenses miss repeat fraud

Today, individual institutions see only the attempts within their own systems, so one agent opening accounts at several banks in a single afternoon can look like a string of unrelated applicants. A fraudster turned away at one institution tries the next with the same face and a new name, and the second institution has no record of the first attempt.

Institutions have long understood that coordinated intelligence would help them combat fraud. In a 2026 Federal Reserve Financial Services survey of 403 U.S. institutions, just over half said they identify mule accounts when another institution alerts them, yet fewer than one in five regularly disclose fraud information.

Sharing at scale has meant pooling or exchanging customer records, which can clash with privacy rules, create a single target for attackers, and risk exposing one institution's customer data to another. As a result, repeat fraud patterns remain hidden.

Introducing TrustGraph

TrustGraph changes how institutions detect repeat fraud. It is the privacy-preserving network that enables banks, fintechs, and other institutions to detect repeat and AI-driven fraud through network intelligence, using patented privacy-preserving technology that keeps every customer record where it was collected. Institutions can learn whether a person or pattern has been tied to fraud at another participating institution without pooling, exchanging, or exposing customer data.

TrustGraph adds network intelligence to the identity decisions institutions already make. The network provides the signal, and each institution decides how to act on it under its own rules. It is not a blacklist.

Three things set TrustGraph apart:

1. Anchored to verified identities. Connect repeat attempts to a verified identity, even when names, documents, or devices change. Devices can be rotated and identity details can be bought, but the verified identity reveals when the same person returns.
2. Intelligence without exposing customer data. Participating institutions get network intelligence without pooling raw customer data in a central repository.
3. Recognize legitimate customers. TrustGraph reveals repeat fraud and recognizes customers already trusted elsewhere, with reason-coded signals that show both risk and legitimacy.

How TrustGraph protects customer data

TrustGraph is built with privacy by design. Patented peer-to-peer cryptography connects fraud and trust signals across participating institutions without creating a central repository of customer records. With TrustGraph, customer data stays within each institution's environment. Signals are de-identified at the source. Raw customer data and images stay with the institution, and matching compares cryptographic shares that do not reveal the underlying data on their own. Secure multi-party computation (SMPC) then compares one-time cryptographic material across participating institutions. SMPC is a cryptographic method that lets several parties learn whether their inputs match without any party seeing another's data, so neither other participants nor Incode can see another institution's data.

The network is peer-to-peer, so there is no central pool of customer records to breach and no central data lake or data broker for attackers to target.

Four privacy protections apply to every query:

• End-user de-identification. Identifiers are de-identified before leaving the institution's environment. Raw customer data does not cross the network.
• Requestor de-identification. Other participants cannot identify which institution submitted the query.
• Voucher de-identification. The requesting institution cannot see which participants contributed to the result.
• No response disclosure. A contributing institution does not see the query outcome.

These protections are built into TrustGraph's architecture, which runs on patented privacy-preserving technology, backed by more than a decade of research conducted by Identiq, a pioneer in cryptographic peer-to-peer identity validation founded by security, privacy, and identity veterans. Identiq was acquired by Incode in June 2026. More than $50 million has been invested in the patented privacy-preserving technology behind TrustGraph.

How TrustGraph works inside a verification attempt

A TrustGraph check runs in four steps:

1. Verify the identity. The customer completes the Incode identity journey the institution already runs.
2. Query TrustGraph privately. Identifiers are de-identified within the institution's environment before TrustGraph queries the network for relevant signals.
3. Receive an explainable signal. TrustGraph returns a category-level result with supporting context, where applicable. Customer records are not returned or exposed.
4. The institution decides. The institution incorporates the signal into its existing decisioning. TrustGraph provides the intelligence, and the institution stays in control.

The query runs in real time, so institutions can act on the result mid-session, whether at onboarding, at login, or during a high-risk transaction.

Up to 53% more fraud detected in early access

In Incode's early access program, TrustGraph surfaced up to 53 additional fraud cases for every 100 detected by an institution's existing models.

Interac, Canada's payments and identity network connecting nearly 300 financial institutions, sees the value of this network approach.

"How do you expect every single business to keep up with deepfakes, synthetic identities, and romance scams on its own? It is impossible. That is why we chose Incode, which has the strongest solution in the market today," said Navjeet Grewal, Managing Director, Verification and Digital Credentials at Interac. "TrustGraph allows institutions to benefit from fraud signals across the network without exposing customer data. If Company A experiences an attack and the same actor later targets Company B, Company B doesn't have to face that attack as if it were the first time," Grewal added.

How TrustGraph recognizes legitimate customers

TrustGraph also recognizes customers who have been successfully verified elsewhere, helping institutions reduce unnecessary friction for legitimate customers. A customer who proves who they are at one institution often arrives at another and starts from zero. TrustGraph gives the next institution a positive signal it can incorporate into its existing risk and decisioning process.

Positive recognition is only returned when the network's requirements for a trusted signal are met, and fraud signals take precedence over trust signals.

Get started with TrustGraph

TrustGraph is available today as part of the Incode platform and works alongside the identity verification, onboarding, and authentication processes that institutions already run. As more institutions participate, the network can recognize more repeat patterns across institutions.

Fraud is already networked. Now your defense is too.

Ready to see how TrustGraph and identity verification run together in a single onboarding session? Request a demo today.