Financial crime is spiking in Singapore. Banks need automation that stands up to auditors
Financial crime in Singapore has escalated to unprecedented levels. Scams have cost more than S$3.4 billion since 2019, according to the Straits Times. In H1 2025 alone, nearly S$500 million was lost across close to 20,000 cases, with government official impersonation scams almost tripling year over year. The trend kept rolling into 2026, including a February variant that tricked victims into handing over gold, jewelry, and luxury watches, totaling at least S$2.9 million in losses.
Digitization is pouring fuel on the fire. Singapore's digital economy added S$128.1 billion in 2024 (18.6% of GDP), with near-universal SME adoption of sector-specific digital tools. Deepfakes are now in the mix: a finance director paid US$499,000 after a Zoom call with AI-generated executives, and surveys show 56% of Singaporean businesses have faced audio deepfakes and 52% have seen video deepfakes.
Agentforce: AI agents built for AML/CFT, risk, and regulatory work
According to a joint paper by Salesforce and PwC, Salesforce's Agentforce gives financial institutions AI-driven capabilities to automate AML/CFT obligations, sharpen client risk profiling, and streamline regulatory filings. The idea is simple: let digital labor handle repetitive, knowledge-heavy tasks so your teams can focus on decisions that move risk and revenue.
Agentforce is an enterprise-grade AI agent platform to build, test, deploy, manage, and orchestrate AI agents across the full lifecycle. It supports low-code and pro-code workflows for both business users and developers. Agents can act on behalf of customers, suppliers, and employees 24/7.
Beyond compliance, agents can handle sales tasks like generating quotes, updating opportunities, surfacing deal context, and coaching reps in real time. In service operations, they resolve routine inquiries, pull data from CRM, and hand off complex cases to humans-shortening response times while reducing errors.
Governance you can defend: MuleSoft Agent Fabric
For fraud detection and compliance, Agentforce integrates with MuleSoft Agent Fabric, a governance and orchestration layer that connects multiple AI agents into one coordinated system. Agents share signals, trigger joint investigations, and reduce time-to-detect.
MuleSoft Agent Fabric enforces security, privacy, and regulatory controls across all agents. That gives banks confidence to scale AI while keeping every autonomous decision safe, compliant, and auditable. Its Agent Visualizer records every agent-to-agent interaction and decision, creating a clear trail: what happened, who did it, and which data was accessed-ready to show your regulator.
Three high-impact use cases for Singapore banks
KYC and CDD automation
Use Retrieval-Augmented Generation (RAG) to ingest and validate client documents, detect expirations, check completeness, and apply onboarding checklists automatically. Centralize logic once; apply consistently across markets and booking centers.
- Shorter onboarding time with less friction
- Higher accuracy and consistency across jurisdictions
- Lean middle-office operations, scalable for managed service environments
Risk scoring automation
Aggregate and analyze internal and external data using configurable rules plus AI-driven insights. Dynamically adjust due diligence levels as risk indicators shift-transaction patterns, geography, adverse media-and trigger escalations when thresholds are breached.
- Real-time risk assessment at account and portfolio levels
- Fewer false positives and lower manual review hours
- Smarter resource allocation to the highest-risk cases
Regulatory compliance management
Automate preparation, validation, and submission of mandatory filings. Extract key data points, validate against current requirements, and generate submission-ready documents with full audit trails.
- Up to 30% reduction in manual data analysis
- Lower error rates and fewer missed deadlines
- Higher confidence meeting local and global standards
Deepfakes raise the bar for controls
Audio and video deepfakes are no longer hypothetical. Your playbook needs out-of-band verification for high-value approvals, biometric liveness checks, and watermarking reviews for media used in authentication. Train agents to flag unusual payment requests, out-of-pattern communication channels, and altered beneficiary data in real time.
What finance leaders should do next
- Map AML/CFT workflows end-to-end. Circle repetitive, knowledge-heavy steps-document intake, screening reviews, case triage, list management, filing prep.
- Stand up an Agentforce pilot for one use case (e.g., CDD refresh). Define KPIs: onboarding turnaround time, false-positive rate, analyst hours per case, STR/SAR timeliness.
- Connect to core systems early. Use MuleSoft Agent Fabric to set data access policies, escalation paths, and audit requirements upfront.
- Harden verification for deepfake risk. Require secondary authentication for large transfers and senior-level approvals. Log every decision path for audit.
- Align with MAS requirements and keep evidence ready. Build reports regulators can walk through easily with Agent Visualizer logs.
Compliance context and references
For policy alignment, review the Monetary Authority of Singapore's AML/CFT notice for banks. It's the standard your controls and audits should be built against.
- MAS Notice 626: Prevention of Money Laundering and Countering the Financing of Terrorism (Banks)
- Singapore Police Force: Scams and Cybercrime Advisories
Further learning
Bottom line
Financial crime is accelerating, and manual controls can't keep pace. Agentforce, paired with MuleSoft Agent Fabric, gives banks a way to automate core AML/CFT work, keep risk scores current, and prove every decision. Start small, measure hard outcomes, and scale what works.
Your membership also unlocks:
