Fraud is no longer just a financial crime, it has become a technological arms race. In 2025, global fraud attacks climbed 8-percent, synthetic identity fraud grew eightfold, and agentic AI traffic surged 450-percent. To make sense of these alarming trends, Enterprise IT News sent questions to Marlene Vicaire, Head of Fraud & Identity for APAC at LexisNexis Risk Solutions, whose team tracks billions of digital interactions across the globe.
EITN: The report reveals an 8-percent rise in global fraud attacks in 2025. What does this tell us about the overall trajectory of cybercrime, are we losing the battle, or is this a manageable escalation?
Marlene: The global increase reflects acceleration rather than failure. Fraud is growing faster than digital transactions again, but this increase follows a relatively stable period in 2024. In parallel, regulatory responses across key regions have tightened digital identity and KYC expectations and increased scrutiny of account takeover, mule activity and cross-border fraud flows. These measures are raising the baseline for compliance and resilience, even as fraud techniques evolve.
At the same time, the data shows that defensive investment makes a measurable difference. Mobile app attack rates declined in most regions, including APAC, showing that focused improvements in authentication, device intelligence and in-app controls can materially reduce risk. This points to a manageable escalation in fraud risk, provided organisations evolve their defences as quickly as criminals evolve tactics.
The data shows fraud is becoming more volatile, driven by automation and coordinated attacks. This does not mean fraud is easier to spot at the individual account or transaction level. Instead, risk is increasingly revealed over time and at scale, where reused infrastructure, abnormal behaviour patterns and cross-platform linkages become visible across systems. This shift means static rules are no longer sufficient; success now depends on adaptive, behaviour- and network-based controls.
EITN: Synthetic identity fraud has grown eight-fold in a single year globally. What specifically changed in 2025 that made this tactic so suddenly attractive to fraudsters at scale?
Marlene: APAC has become a prime environment for synthetic identity fraud because of rapid digital onboarding, large addressable populations and cross-border platform use. First, synthetic identities became easier to industrialise. Fraudsters increasingly combine stolen data fragments with generative AI to construct identities that look valid, consistent and persistent over time.
Second, these attacks exploit a structural weakness: there is no real victim to raise an alert. As a result, synthetic identity fraud can remain undetected for long periods, especially during account creation and early account activity. That combination of scalability, longevity and low immediate risk made synthetic identities especially attractive at scale across APAC and globally.
EITN: Traditional fraud detection often relies on finding a victim who raises the alarm. With synthetic identities, there is no real victim to trigger that alert. How fundamentally does that change the detection and prevention model organisations need to adopt?
Marlene: This is an outdated view. Reactive approaches have long been abandoned and shifted to preventative detection. The further upstream organisations act, the more likely they are to stop activity before it is too late.
Without a victim, organisations can no longer depend on post-transaction complaints or chargebacks. Detection must happen before value is extracted, using behavioural signals, device intelligence and network-level pattern recognition. The report shows that isolated, transaction-level checks are insufficient. Effective defence increasingly depends on network-level intelligence, behavioural intelligence and cross-platform visibility.
For example, geographically clustered activity, repeated infrastructure reuse or coordinated behaviour across multiple accounts may appear legitimate in isolation but reveal organised fraud when viewed collectively.
This is especially critical in APAC, where fraud networks often span multiple countries and industries.
EITN: Fraudsters are reportedly using generative AI to create synthetic identities complete with fraudulent historical backup data. How do you detect an identity that has been deliberately engineered to look authentic over time?
Marlene: Detection is less about spotting ‘fake’ profiles and more about identifying inconsistencies across behaviour at scale. While synthetic identities may appear internally coherent, they often fail when examined across devices, networks, geographies and timelines.
As agentic systems become more autonomous and human-like, organisations will need network-level understanding of behaviour, rather than relying on traditional bot detection.
Marlene Vicaire
The LexisNexis Digital Identity Network shows that engineered identities tend to reuse infrastructure like device and email addresses, exhibit unnatural interaction patterns, or link back to known fraud networks over time. Detection therefore depends less on individual data points and more on long-term pattern analysis and cross-industry intelligence.
EITN: Agentic traffic grew 450-percent from Q1 to Q4 2025, yet there is no current indication of malicious intent. How do you draw the line between a legitimate AI shopping agent and a malicious one, and how long before that distinction becomes critically difficult to make?
Marlene: Today, the distinction is still possible by analysing context, behaviour and intent — not simply automation. For example, how transactions are sequenced, how identities evolve and how infrastructure is reused. Legitimate agents tend to behave consistently within user context, while malicious automation optimises for scale, speed and evasion.
However, that line is narrowing. As agentic systems become more autonomous and human-like, organisations will need network-level understanding of behaviour, rather than relying on traditional bot detection. This transition is approaching quickly in digitally advanced APAC markets. That is why maintaining visibility across transactions, channels and industries becomes increasingly critical.
EITN: If current trends continue unchecked – synthetic identity fraud growing eight-fold, agentic bots scaling 450-percent – what does the fraud landscape realistically look like in 2027?
Marlene: By 2027, fraud will be less about individual attacks and more about competing systems. We would expect highly automated, cross-border networks that continuously test defences, adapt in near real-time and exploit gaps between industries and regions.
At the same time, the report shows that collaboration, shared intelligence and network-based detection are beginning to deliver results. The future is not predetermined. In APAC especially, the outcome will depend on whether defences remain fragmented, or evolve into coordinated, ecosystem-wide responses. The organisations that invest in shared visibility and adaptive controls will bend the curve; those relying on static models will struggle.