Feedzai launches RiskFM foundation model for financial crime prevention

Feedzai has introduced RiskFM, a foundation model designed for financial crime prevention and built specifically for tabular financial data. The company describes it as the first tabular foundation model developed for risk decisioning across fraud, anti-money laundering and related financial crime use cases.

The launch reflects a broader effort to move beyond traditional fraud systems built on rules and manually engineered models for individual institutions. Feedzai says RiskFM is designed to operate across the full financial crime lifecycle, covering areas such as onboarding, digital activity, payments, transfers and AML workflows. The aim is to give financial institutions a model that can be deployed faster, adapted more easily and trained across broader datasets than conventional approaches.

A central challenge, according to Feedzai, is that financial transaction data behaves very differently from language, audio or image data. Transactions are less predictable, patterns shift continuously and the domain is adversarial, with fraudsters constantly adapting their methods. RiskFM is intended to address that complexity by learning from large-scale, multi-dimensional financial datasets rather than relying on narrow or highly siloed models.

Feedzai says the model has been tested across data spanning trillions of dollars in payments and billions of events worldwide. In its early results, the company claims RiskFM can match the performance of highly tuned supervised models when deployed for a single institution, while outperforming traditional approaches when trained across multiple institutions and geographies. That, in turn, could reduce implementation time, lower maintenance costs and improve adaptability across a wider set of risk use cases.

The company plans to continue validating the model with early adopters before extending the framework across its wider product suite.

About the Company

Feedzai is a financial risk management and crime prevention technology company focused on fraud detection, anti-money laundering and risk decisioning. The firm works with financial institutions globally and positions itself around AI-driven tools for monitoring transactions and detecting suspicious activity across the financial services landscape.

Why This Matters to FinanceX Readers

Financial institutions are under pressure to modernise fraud and AML systems without creating more operational complexity. Feedzai’s launch matters because it signals an attempt to bring foundation model architecture into one of the most difficult areas of financial services: fast-moving, adversarial transaction monitoring.

For banks, payment providers and compliance leaders, the key question is whether these models can deliver stronger detection rates and lower deployment costs without sacrificing transparency or control. That makes RiskFM a development worth watching.