Last month, Oak portfolio company Feedzai released their 2025 AI Trends in Fraud and Financial Crime Prevention report, which offers a comprehensive snapshot of how financial institutions are leveraging advanced AI to stop increasingly sophisticated fraud attacks. The report takes a deep look at how Al is confronting the complex ethical, operational, and regulatory challenges that it faces.
Feedzai’s report finds that 90% of financial institutions are now using AI to fight fraud and financial crime, with 64% having adopted AI within the past two years. As AI becomes a pillar of modern risk management, it’s clear that success hinges on more than just adoption—it requires trustworthy, transparent, and collaborative systems.
One of the most urgent themes in the report is the rise of generative AI-enabled fraud. 60% of surveyed professionals say criminals are using GenAI for voice cloning scams, while others cite deepfakes, phishing, and AI-written social engineering content. Feedzai warns that these capabilities lower the barrier to entry for bad actors and increase the scale and effectiveness of attacks.
Based on the report findings, companies are not entirely unprepared for this emerging threat. Nearly all (96%) of AI-using financial institutions are themselves deploying GenAI to strengthen detection, streamline investigations, and analyze anomalous behaviors in real time. Feedzai's RiskOps platform analyzes a customer's behavior over their full life cycle of interactions – from account opening to ongoing account activity, behavioral data, and transaction monitoring. This compares to traditional fraud or transaction monitoring techniques that are mostly focused on single events, which are now more easily broken by GenAI fraudsters. For example, scams, which are often transactions by an authorized user, have been one of the fastest rising fraud techniques scaled by GenAI. They have caused $1T in fraud in the last year and 50% of consumers encounter them once a week. Feedzai's ability to detect anomalies over the full lifecycle of customer touch points helps prevent these more complex scams, money mule accounts, and other emerging fraud behaviors that GenAI has scaled.
Data quality and management also emerged as a top concern, with 87% of respondents citing it as a key challenge, particularly as the volume of digital transactions skyrockets. Additionally, 59% of institutions that haven’t yet adopted AI say concerns over data availability and integrity are their biggest barrier. Data quality and security are top of mind because they are pre-requisites to data interoperability, one of the most valuable promises of AI. Models and tools get more powerful when they consume data from more sources. To date, the bottleneck has been the tedious time it takes to merge data taxonomies that were designed for independent systems. In contrast, AI can write its own thesaurus, which means independent datasets are interoperable by default.
In the report, the Feedzai team also lays out their TRUST Framework (Transparent, Robust, Unbiased, Secure, Tested) to provide a roadmap for how AI should be designed and deployed across industries. From federated learning to bias mitigation and explainability, the report calls for proactive governance as a core component of any fraud-fighting AI stack. For example, Feedzai IQ's fraud intelligence platform uses data from 100 clients, 8T in annual payment volume, and 70B events to build federated fraud models. Prove, another Oak HC/FT portfolio company, has built an identity network and trust score based on 15 years of historical data and 30b+ annual events from its 1,000+ customers to green flag or red flag mobile numbers, devices, and identities.
Feedzai uses the report as an opportunity to challenge businesses to follow the playbook of bad actors—who readily share tactics and collaborate—by building their own coalitions for collective defense. From consortium data sharing to adversarial testing and AI “red teaming,” the report outlines how institutions can collaborate more effectively while protecting privacy and proprietary systems.
Importantly, the Feedzai survey reveals that while AI is reshaping operational dynamics, it’s positioned to only enhance, not replace human decision-making. At Oak HC/FT, we’re proud to partner with companies like Feedzai that are not only advancing technical excellence but also leading the charge for ethical, transparent, and impactful AI. As the threat landscape evolves, the future of fraud prevention will depend on those willing to invest in both innovation and integrity.