As artificial intelligence gives fraudsters new ways to impersonate trusted institutions and manipulate consumers, payment security is moving beyond protecting transactions to protecting the decisions that happen before a payment is made. Visa is responding with a new security strategy for Singapore that puts AI-driven fraud detection, stronger authentication, tokenisation and ecosystem-wide intelligence sharing at the center of its next phase of payments security.

Visa Targets AI-Driven Scams With New Singapore Payments Security Roadmap

Visa has unveiled its Singapore Security Roadmap 2026 and Beyond, outlining six priorities intended to strengthen the country’s digital payments infrastructure against increasingly sophisticated fraud and scams.

The roadmap focuses on cybersecurity, authentication, tokenisation, e-commerce checkout, payment-security standards and broader ecosystem resilience. The common thread is a shift from treating fraud as a transaction-level problem to addressing threats across the entire payment journey.

That distinction is becoming increasingly important as AI changes both sides of the payments equation. Financial institutions and payment networks can use artificial intelligence to identify suspicious activity at scale, but criminals can also use automation, synthetic identities, impersonation and social engineering to make scams more convincing.

For Singapore, where digital payments are deeply embedded in everyday commerce, the challenge is maintaining consumer confidence without adding unnecessary friction.

The fraud problem is moving upstream

Singapore’s latest scam data illustrates why payment providers are increasingly focusing on authorised transactions and social engineering rather than only unauthorised card fraud.

The Singapore Police Force reported 37,308 scam cases and approximately S$913.1 million in losses in 2025. While both figures declined substantially from 2024, the scale of financial losses remains significant. Cryptocurrency-related losses alone accounted for about S$182.2 million, or 20% of total scam losses.

More importantly, many scams do not require criminals to break into a victim’s account.

Singapore Police data shows that self-effected transfers accounted for 81.8% of scam cases in 2025, meaning victims were manipulated into making the transactions themselves.

That creates a difficult problem for conventional fraud systems. A legitimate customer may be authenticated correctly, using a genuine device and valid credentials, while still transferring money because a scammer has successfully manipulated them.

Visa’s roadmap therefore puts greater emphasis on technologies that can assess risk throughout the payment experience.

Six priorities for a more resilient payments ecosystem

Visa’s updated strategy identifies six areas for investment.

The first is cybersecurity, with an emphasis on anticipating emerging threats rather than responding only after an attack. The second is authentication, including approaches designed to provide stronger verification without making digital payments cumbersome.

The third priority is tokenisation. Instead of repeatedly exposing sensitive payment credentials, tokenisation substitutes payment information with a secure digital token. Visa has increasingly positioned tokenisation as a core component of digital commerce security, particularly as payments move into mobile wallets, connected devices and emerging AI-driven commerce environments.

The fourth priority addresses e-commerce checkout. Visa wants security measures to become less visible to legitimate customers while still giving payment providers stronger signals for identifying suspicious activity.

The remaining priorities focus on foundational standards and risk programs, followed by broader ecosystem resilience.

That last point may prove the most consequential.

Fraud rarely stops at the boundary of one company. A scam can begin on a social platform, move through a messaging service, involve a merchant, trigger a bank transfer and eventually reach a payment network or cryptocurrency platform.

No single participant has complete visibility.

AI is becoming both a defensive tool and an attack surface

Visa says it currently deploys more than 150 AI and machine-learning models globally to identify fraud, optimise authentication and protect digital payments.

The company is also expanding AI-powered capabilities designed to detect scams before they become large-scale operations. Visa says its Scam Disruption program uses AI to map scam infrastructure, analyse transaction anomalies and share intelligence across markets.

That matters because AI-enabled fraud can operate at a different scale from traditional criminal activity.

Generative AI can make phishing messages more convincing. Automated systems can personalise communications. Deepfake technologies can make impersonation attacks harder to distinguish from legitimate interactions. Visa has described the emerging threat as a shift toward attacks on identity and trust rather than simply stolen payment credentials.

For payments companies, this changes where security controls need to operate.

A fraud engine that only evaluates the final transaction may miss the signals generated during account creation, authentication, browsing, communication or checkout.

Tokenisation and authentication become infrastructure, not features

Visa’s approach also reflects a broader transformation in payment infrastructure.

Tokenisation, passkeys and biometric authentication are increasingly being positioned as foundational technologies rather than optional security features. Visa says tokenised transactions have materially lower fraud rates than traditional PAN-based transactions, while biometric authentication can significantly reduce fraud compared with one-time passwords.

For banks, fintechs and merchants, the implication is that security architecture will increasingly need to combine several layers: identity verification, device intelligence, behavioural signals, tokenisation, authentication and transaction monitoring.

The objective is not necessarily to add more authentication prompts. It is to make risk assessment more intelligent and largely invisible when the transaction appears legitimate.

Collaboration becomes part of the security stack

Visa’s roadmap also highlights collaboration with regulators, banks, merchants, fintechs and other industry participants.

That reflects a reality that is becoming difficult for the payments industry to ignore: fraud intelligence is most valuable when it can move across organisational boundaries quickly.

Singapore has already demonstrated the value of coordinated intervention. In 2025, the Singapore Police Force said its Anti-Scam Command and partners recovered more than S$140 million in scam proceeds and helped avert at least S$348 million in potential losses through proactive interventions.

The next challenge is making those interventions increasingly real-time and data-driven.

For enterprise payments teams, Visa’s roadmap points toward a security model in which fraud prevention is no longer owned exclusively by the payments or cybersecurity department. Product teams, identity specialists, merchants, banks, fintechs and regulators will increasingly need shared standards and interoperable risk signals.

That may ultimately determine whether AI becomes a net security advantage for digital commerce—or simply gives both attackers and defenders more powerful tools.

Market Landscape

Singapore’s payments market sits at the intersection of several major fintech trends: real-time payments, mobile commerce, tokenisation, digital identity, open banking and AI-driven fraud prevention.

The country’s scam environment also demonstrates why authorised payment fraud is becoming an industry priority. The Singapore Police Force reported that investment scams generated S$336.2 million in losses in 2025, while government-official impersonation scams generated another S$242.9 million.

The competitive landscape increasingly includes payment networks such as Visa and Mastercard, banks such as DBS and OCBC, fintech platforms, digital-wallet providers and technology companies building identity and fraud infrastructure.

Visa’s strategy also fits a broader payments-industry movement toward AI-native commerce. As AI agents eventually begin searching for products, comparing offers and initiating purchases, payment providers will need to determine not only whether a transaction is legitimate, but whether the entity initiating it is authorised to act for the consumer.

That makes authentication, tokenisation and identity infrastructure strategically important well beyond conventional card payments.

For enterprise teams, the takeaway is clear: fraud prevention is becoming an end-to-end infrastructure problem rather than a point solution.

Top Insights

• Visa’s six-part Singapore roadmap targets cybersecurity, authentication and tokenisation as AI makes payment scams more personalised and difficult for conventional fraud systems to detect.
• Singapore recorded S$913.1 million in scam losses during 2025, underscoring why banks, merchants, fintechs and payment networks need stronger shared fraud intelligence.
• Self-effected transfers represented 81.8% of Singapore’s 2025 scam cases, shifting security priorities from stolen credentials toward behavioural and social-engineering detection.
• Visa’s 150-plus AI and machine-learning models demonstrate how payment networks are embedding artificial intelligence directly into fraud detection and authentication infrastructure.
• Tokenisation, biometrics, passkeys and ecosystem-wide intelligence sharing could become foundational controls as digital commerce moves toward increasingly automated and agentic payments.

Get in touch with our fintech expert