The rapid evolution of AI-driven fraud has been undeniably forcing enterprise finance leaders across APAC to rethink how they protect increasingly digital and interconnected payment operations.
According to a study by Juniper Research, fraudulent transactions in digital banking and money transfer are projected to surpass 2.2 billion globally by 2031, up more than 180% from 2025, reflecting a threat that could have direct implications for CFOs managing treasury, payments and financial controls.
As fraudsters increasingly target customers and authenticated payment journeys rather than banking infrastructure itself, the findings point to a growing need for CFOs to reassess whether their organisations’ existing controls can keep pace with increasingly personalised and adaptive attacks.
Shane O’Sullivan, research analyst at Juniper Research, believes that CFO’s and finance leaders in Asia Pacific must reassess fraud as a broader financial and operational risk, rather than focusing only on direct losses.
“The region is seeing some of the fastest growth in fraudulent transactions and digital payments, creating far larger and more complex data environments across e-wallets, instant payments and increasingly interconnected domestic and cross-border payment networks.”
O’Sullivan thinks the key issue is whether institutions can turn that data into a complete view of risk.
“Fragmented legacy systems and siloed infrastructure can leave fraud teams with incomplete visibility across customers, channels and markets, while criminals increasingly exploit those gaps through Authorised Push Payment fraud, synthetic identities and mule networks.”
He says finance leaders should therefore prioritise the quality, consistency, and connectivity of fraud data, not simply the volume collected, so institutions can identify relationships and anomalies across the wider transaction journey.
“Investment in new fraud technology will only deliver value if the underlying data is accurate, timely and integrated.”
For APAC institutions, O’Sullivan says the opportunity is to avoid embedding the fragmented fraud infrastructure that more established banking markets are now having to modernise.
“As fraud becomes increasingly adaptive, the competitive advantage will not necessarily belong to the institution with the most data, but the one that can turn the highest quality data into a complete and timely view of risk.”
Investments in fraud prevention
As generative and agentic AI make fraud more personalised and adaptive, finance leaders are faced with the task to prioritise investments in fraud prevention, transaction monitoring, and AI-driven risk management, and assess the ROI of those investments.
O’Sullivan says finance leaders should prioritise investment that improves decision quality across the full fraud lifecycle, including behavioural analytics, identity intelligence, graph-based detection, adaptive transaction monitoring and AI-assisted investigation.
He adds that agentic AI can help automate alert triage and case management, but it should augment rather than replace human oversight.
“Return on investment should be measured beyond fraud losses prevented,” he opines. “CFOs should assess reduction in false positives, manual review costs and investigation time, alongside improvements in detection accuracy, and customer friction.”
“The strongest investments are those that improve both fraud outcomes and operational efficiency.”
A shift
As the report suggests banks need to move beyond transaction-level detection towards a continuous view of customer risk, CFOs must weigh in on the transformation in terms of data governance, financial controls and collaboration between finance, treasury, risk and technology teams.
“A continuous view of customer risk requires banks to connect identity, behavioural, account, and transaction data across the customer lifecycle rather than assessing individual payments in isolation.”
For CFOs, O’Sullivan says this increases the importance of clear data ownership, consistent risk definitions, and strong governance over how data is collected, shared and used in automated decisions.
“It also requires closer collaboration between finance, treasury, fraud risk and technology teams, as fraud exposure can no longer be managed within a single function. Shared data and risk indicators should allow these teams to respond to changes in customer behaviour before they translate into financial loss.”
Adapting to tactics
As fraudsters use AI to adapt their tactics in real time, APAC finance leaders should factor increasingly dynamic fraud risks into enterprise risk assessments, budgeting and business continuity planning.
For O’Sullivan, finance leaders must look to move away from treating fraud risk as a largely static annual assumption.
“AI is increasing the speed, scale and sophistication of attacks, meaning enterprise risk assessments should incorporate more severe and rapidly evolving fraud scenarios.”
He adds that INTERPOL has warned that criminals across APAC are increasingly using AI and sophisticated social engineering at scale, with agentic AI as an addition to the fraudsters arsenal this issue is only set to compound.
“CFOs should therefore build greater flexibility into fraud budgets, stress test exposure to sudden changes in attack patterns and ensure funding can be redirected quickly when new threats emerge.”
He says business continuity planning should also consider fraud-model failure, third-party technology disruption and situations where disruption and situations where existing controls become temporarily ineffective; thus, selecting banking fraud prevention vendors with high uptime becomes increasingly important.
“The objective is to build financial resilience around an evolving threat, rather than budgeting against a fixed fraud rate.”
Given the growing use of AI on both sides of the fraud equation, and to ensure that fraud-prevention capabilities can keep pace with emerging AI-enabled threats, O’Sullivan believes CFOs should ask their banking and technology partners the questions:
How quickly can your fraud models detect and adapt to new AI-enabled attack patterns, and how often are retrained and tested?
If AI or agentic AI is being used to support investigations, how transparent are its decisions, and can investigators clearly understand and challenge the reasoning behind them?
How effectively can your systems identify coordinated fraud across accounts, channels, and institutions rather than assessing suspicious activity in isolation?
Looking ahead
Looking ahead over the next 12–24 months, CFOs and finance leaders in APAC are in need to prepare their organisations for the growing sophistication of AI-enabled fraud, considering the areas of investment or financial controls to be prioritised at present.
O’Sullivan says finance leaders should prepare for AI to increase both the sophistication and scalability of fraud.
“Agentic AI is particularly important because it could allow criminals to automate multiple stages of an attack, while deepfakes, synthetic identities and AI-generated social engineering are already making impersonation considerably more convincing.”
He notes that investment should therefore focus on adaptive fraud models, stronger identity and deepfake detection, and network-level capabilities that can identify mule accounts and coordinated fraud across multiple customers and institutions.
“CFOs should prioritise clear governance around the use of artificial intelligence, including explainability, human oversight and contingency controls where automated models fail.
“The objective should therefore be to build fraud controls that can adapt as quickly as the underlying threat.”
