AI fraud prevention for banks: reducing scam losses with shared intelligence

AI scam prevention and banking fraud protection with a digital payment card and warning symbols

Article | September 16, 2026

Banks can reduce AI scam losses with shared intelligence

A customer doesn’t wake up one day, intent on sending their life savings to a criminal.

The fateful decision is shaped earlier, across channels their bank doesn’t own: a cloned voice, a fake investment group, a recruitment scam, a manipulated marketplace exchange and so on.

That’s why AI-enabled scams demand a new executive response. Banks remain central to protecting money, but they can’t be the only line of defense when the decisive moment of fraud often happens before a payment is made.

The scam boundary has moved beyond the bank

Traditional fraud strategies focus on credentials, transactions and account behavior. Those controls still matter. Yet the fraud journey has expanded. Criminals can build trust on social platforms, move victims to messaging apps, apply pressure by phone, create synthetic documents and then use instant payments, crypto transfers or mule accounts to extract funds.

AI makes this journey faster and more convincing.

It can personalize language, generate plausible identities and create messages that don’t look like the clumsy phishing attempts customers were taught to spot. The risk spans a broader trust infrastructure that includes banks, telecoms, social platforms and marketplaces.

The FTC said that U.S. consumers reported more than $12.5 billion in fraud losses in 2024 (up 25% from the previous year), while the FBI reported more than $16 billion in losses from suspected internet crime.

UK Finance reported that criminals stole £1.17 billion through unauthorized and authorized fraud in 2024, while banks prevented £1.45 billion of unauthorized fraud. It also found that 70% of authorized push-payment fraud cases began online, and 16% began through telco networks.

AI changes the economics of manipulation

A scam succeeds when the victim believes the story, feels urgency and trusts the instruction. AI helps criminals enhance each part of this process. A fake job offer can be accompanied by credible documentation. A corporate email compromise attempt can imitate a senior executive. A voice note can appear to come from someone that the customer knows.

For banking leaders, the implication is clear: fraud prevention can’t depend on customers recognizing weak signals. FS-ISAC’s 2025 outlook identifies GenAI-enabled fraud and scams as a key threat to testing financial-sector resilience and calls for fraud and cyber teams to share intelligence more effectively.

Time for a collective industry response

Banks can detect changes in device behavior, payment patterns, beneficiary details and account activity. But they often don’t know that the customer has spent weeks in a fake investment group, received calls from an impersonator or been coached to ignore bank warnings.The industry needs a wider operating model for scam prevention.

Banks must continue improving payment analytics, identity controls, behavioral insight and customer education. At the same time, earlier signals need to flow from the channels where scams originate:

Telco providers can help to identify spoofing and suspicious call patterns.
Digital platforms can act more quickly against impersonation, fraudulent advertising and fake investment communities.
Payment firms can detect mule behavior.
Regulators can clarify how intelligence can be shared safely across sectors.
Law enforcement can use better reporting to disrupt groups.

Adopt a practical agenda

The International Monetary Fund (IMF) has warned that financial cyber risk is increasingly cutting across shared infrastructure, affecting confidence, critical services and stability. Those same ideas apply to AI scams: a single-institution response is too narrow for networked crime.

The practical agenda for banks to deal with this begins with a more detailed map of the scam journey. Decision-makers should determine where customers are first contacted, how criminals build credibility, which channels are most abused and where the bank can intervene without unnecessary friction.

Banks also need to create more targeted warnings. A generic “are you sure you want to make this transaction?” message has little value when a customer has been coached by a criminal. Better interventions should reflect the type of payment, customer context, beneficiary risk and known scam patterns.

The operating model is as important as the technology. Fraud, cyber, financial crime, customer service and digital product teams need shared intelligence and common escalation routes. Executives need measures beyond loss numbers, including recovery speed, customer harm, complaint volumes and trust.

Real-world examples show how AI, modernization and human oversight can improve banking outcomes.

NatWest Group digitized check clearing, increasing speed and accuracy and enabling automated image fraud detection using machine learning and AI. The program reduced check-clearing costs by 50%, cut clearance cycle time by 66%, processed around 350,000 checks in a 2-hour window with 99.9% accuracy and identified an average of 350 fraudulent checks a month, saving more than £50 million in potential fraud losses each year.

DXC Technology helped deliver the NatWest Group transformation. More broadly, DXC works with more than 350 active financial services clients and 17 of the top 20 global banks, and our Hogan core banking platform supports 300 million deposit accounts and two-thirds of U.S. card transactions.

Our AI capabilities, DXC Xponential and DXC OASIS, are designed for the harder part of this agenda: moving from isolated AI pilots to governed AI at scale, and from fragmented operations to real-time visibility across complex environments. DXC Xponential provides a repeatable blueprint for secure, responsible enterprise AI, while DXC OASIS combines human expertise with agentic AI to support predictive, resilient operations across existing IT estates.

Fighting AI fraud together

Although they remain central to the fight against scams, banks can’t be expected to police every digital interaction that happens before a payment instruction.

Consequently, as AI makes scams more convincing, the industry needs to progress from institution-by-institution fraud control to ecosystem-wide trust protection. Banks must share intelligence, use AI responsibly, strengthen human intervention and treat scam prevention as a core trust strategy.

Leaders who act now will reduce losses, protect customers and heighten confidence in digital finance.