Asia-Pacific digital economies face highest identity fraud

Shufti has published research showing that digitally advanced economies in Asia-Pacific face the highest rates of identity fraud signals, with Indonesia recording the highest rate in its dataset.

The findings suggest that markets with stronger digital adoption and higher volumes of remote onboarding are also seeing more fraudulent attempts during customer verification. Fraud-signal rates per 1,000 verifications ranged from 22.82% in Indonesia to 2.57% in Malaysia. New Zealand, India, Australia, Japan and Singapore also ranked near the top, while Thailand, Vietnam and Bangladesh were at the bottom.

The report focuses on know-your-customer checks across the region, where digital businesses are handling growing numbers of customers remotely as regulators tighten standards for identity checks, liveness detection and fraud prevention.

According to Shufti, the issue is less about weak defences than exposure. More digital-first businesses, more remote onboarding and more verification traffic create more opportunities for criminals to test stolen, synthetic or manipulated identities.

That trend is unfolding in some of the region's largest digital ecosystems. India's UPI now handles roughly half of all real-time payments worldwide, while Singapore's Singpass processes more than 41 million transactions a month, increasing the volume of identity decisions businesses must make in real time.

"What our data makes clear is that fraud scales with digital maturity, not with weak defences," said Ammara Mukhtar, Regional Vice President, Sales APAC, Shufti.

"The most digitally advanced markets in Asia-Pacific are experiencing the highest levels of fraud activity while regulators continue to introduce new compliance requirements at an unprecedented pace."

Regional complexity

The challenge grows as businesses expand into multiple Asia-Pacific markets. Verification volumes can rise from the low hundreds of thousands to around 10 million a year as companies enter three or four markets, while document formats, writing systems and local fraud patterns become harder to manage through a single process.

Shufti highlighted the operational strain created by the region's linguistic and administrative diversity. Identity documents may appear in Latin script, Devanagari, Khmer, Burmese, Thai, Korean Hangul, simplified and traditional Chinese, Japanese scripts and Urdu. Some languages, including Thai and Khmer, are written without spaces between words, complicating efforts to separate names and addresses for screening. Korean names may be romanised in several ways, and Japanese records can combine multiple scripts and use imperial-era dates instead of the Gregorian calendar.

Shufti argues that many existing verification systems were built around Western document standards and Roman alphabets, leaving them poorly suited to Asia-Pacific onboarding requirements. In that view, those technical limits constrain how many markets a business can serve with confidence.

Tighter rules

The regulatory backdrop is also changing. In India, the Reserve Bank of India's updated KYC Master Direction raised expectations for spoof detection and liveness in video-based customer identification. Malaysia's revised e-KYC policy now mandates liveness detection. Singapore's Monetary Authority published an information paper on cyber risks linked to deepfakes. Australia's AML/CTF Amendment Act 2024 extends mandatory identity verification to lawyers, accountants and real-estate professionals. In Vietnam, an administrative restructuring changed provincial boundaries and removed the district tier, creating document and address-matching complications.

Those changes come as fraud methods evolve quickly. Shufti cited United Nations Office on Drugs and Crime data showing a rise of 600% or more in deepfake-related content tied to criminal activity in Southeast Asia in the first half of 2024. It also pointed to the USD $25 million loss in the Hong Kong Arup case, which involved a deepfake video call impersonating senior executives, and to FinCEN estimates that more than USD $4 billion was laundered through the Cambodia-based Huione Group between August 2021 and January 2025.

Fraud is increasingly organised in coordinated waves that can appear independent at the document layer, affecting businesses from fintech groups handling mule accounts to crypto exchanges facing synthetic identity attempts.

Commercial pressure

For businesses, the impact is not confined to crime losses or regulatory penalties. The report argues that onboarding friction can reduce conversion rates as customers abandon identity checks before completing sign-up, creating a revenue drag that is often hard to measure directly.

It also points to the cost of enforcement when controls fail. In Singapore, the 2023 money-laundering case involving more than SGD 3 billion in seized or frozen assets led to penalties totalling S$27.45 million against nine financial institutions in 2025.

Many vendors struggle because they depend on licensed core technology, use optical character recognition systems trained mainly on Western documents, or rely on static fraud models that are not updated quickly enough, according to Shufti. The result is often a patchwork of tools from different suppliers that increases privacy, residency and operational risks without fully closing verification gaps.

"Businesses can no longer rely on fragmented systems and disconnected verification tools to address these challenges. A unified verification platform enables organisations to respond faster to evolving fraud patterns, simplify compliance operations and deliver a consistent customer experience across markets. Success in APAC depends on understanding the region's diverse documents, regulatory requirements and emerging fraud risks while adapting quickly to constant change," Mukhtar said.

Shufti said its system is built in-house rather than assembled from third-party components, and that it covers more than 10,000 document types across more than 240 countries and territories in over 150 languages. It added that the system achieved 96.79% field-level OCR accuracy on Vietnam's new chip-based citizen ID during the country's transition, compared with 82.36% for a comparable third-party benchmark.