TAKEAWAYS
Many organisations treat fraud as a finance, legal and compliance, or internal audit problem. Digital forensic expertise appears only when laptops have to be imaged, documents have to be recovered, or a breach has already happened. This model is now outdated.
Modern financial crime is predicated on digital access, digital trust, and digital speed. A fake supplier payment may start with a compromised email account. A bribery scheme may be coordinated through encrypted messaging. A scam victim may be defrauded through social media, instant messaging, fake investment platforms, cryptocurrency wallets, and funds stolen by account takeover fraud and laundered via mule accounts.
The United States’ Federal Bureau of Investigation’s Internet Crime Complaint Center (IC3) reported more than US$20 billion in losses in its 2025 Internet Crime Report, with investment fraud, business email compromise, and tech support scams among the major loss categories. The numbers are a warning, but what matters more is that fraud is increasingly digital, social-engineered, and fast-moving.
Forensic professionals who only understand ledgers, invoices, approvals, and bank statements are now seeing only part of the story.
Generative AI (GenAI) and agentic AI have emerged as powerful tools for fraudsters. A fraudster can now draft polished emails in multiple languages, clone voices, create fake profile images, produce synthetic documents, generate fake corporate identities, and automate parts of social engineering, all in record speed and with minimal effort. Europol, in its 2025 Serious and Organised Crime Threat Assessment Report, has warned that artificial intelligence (AI) can support attack automation, social engineering, security bypasses, deepfakes, synthetic media, and false identities. The Thales Bad Bot Report reported the average number of AI-driven bot attacks mitigated increased by a staggering 12.5 times in 2025.
People can no longer rely on traditional warning signs to detect scams. The conventional poor grammar, strange formatting, suspicious accents, amateur websites, and clumsy invoices are no longer dependable red flags. The deception enhanced by AI is immaculate. The pressure to commit feels more real. The request may be very convincing and from a seemingly “close acquaintance”.
A reported attempted scam involving a former Chief Financial Officer of a Malaysian bank is a case in point. According to The Edge Malaysia, impersonators allegedly used WhatsApp and a Zoom call in connection with a requested transfer of almost US$1 million. The bank suffered no loss, but the case shows how executive impersonation, video interaction, messaging applications, and payment processes can converge to create highly convincing scams. In Singapore, a person lost at least S$4.9 million after falling for a ruse which included a deepfake AI video conference call with senior government officials and Prime Minister Lawrence Wong.
The lesson is clear – an email, a message, voice or video call, an important document, or a familiar communication channel can no longer be taken at face value. Additional steps, as a default, are now necessary to validate authenticity.
The modern fraud file is no longer made up only of invoices, approvals, emails, bank records, and signed documents. It likely includes login logs, device data, metadata, chat records, cloud access history, cryptocurrency flows, screenshots, and records from third-party platforms.
Cellebrite’s 2022 Industry Trends Survey revealed that most law enforcement agencies and prosecutors believe that digital evidence is more important than physical evidence and DNA in successfully prosecuting cases. The same shift is now visible in corporate investigations – the evidence is increasingly digital, fragmented, fast-moving, and spread across business systems, personal devices, cloud platforms, messaging applications, and external service providers.
Poor digital evidence handling can jeopardise an investigation. A device can be altered. A message can be deleted. A cloud record can expire. A fake website can disappear. A cryptocurrency trail can move through exchanges, mixers, bridges, or mule wallets. A fraudster can leave the organisation before access logs are preserved.
Forensic professionals have to recognise digital threats well enough to ask better questions, preserve the right evidence, and relate technical events to financial impact. This means understanding how phishing leads to payment fraud, how credential theft supports account takeover, and how metadata can confirm or disprove a claim.
As a defensive measure, forensic professionals have a critical role to support organisations to proactively design and embed anti-fraud controls on top of cybersecurity controls into processes and IT systems, before a product, platform, payment workflow, or customer channel goes live.
AI governance frameworks must ensure that fraud risks and controls are rigorously managed for AI solutions implemented within the organisations. A new payment flow should be resilient against business email compromise, account takeover and fake vendor changes. A new onboarding process should be designed to detect synthetic identities, document tampering and device intelligence anomalies.
Singapore’s regulatory direction reflects this shift. The Monetary Authority of Singapore (MAS) and Infocomm Media Development Authority (IMDA)’s Shared Responsibility Framework assigns duties to financial institutions and telecommunication companies to mitigate phishing scams, with payouts to affected scam victims where prescribed duties are breached.
This is not merely a customer service issue after money is lost. It is an expectation that institutions should design, operate, monitor, and implement evidence controls that reduce digital fraud harm.
Organisations should therefore ask whether anti-fraud, cyber, compliance, legal, finance, and internal audit teams are working off the same threat scenarios. Criminals do not operate in organisational silos and neither should the response.
From the cases and examples highlighted above, it is evident that digital trust has become easier to fake and harder to verify. The organisations that respond well to emerging threats by AI and fraud through digital channels will be those that break down operational silos and merge financial crime, forensic investigation, digital forensic and cybersecurity into a unified defence.
In conclusion, boards and executive leaders should recognise cyber-enabled fraud as a strategic business risk and consider the following key actions:
1) Build and test a joint fraud-cyber crisis playbook
Map how cyber events can lead to financial loss, regulatory exposure, customer harm, and evidence loss. Test the playbook through realistic tabletop exercises. A serious cyber-enabled fraud event requires containment, evidence preservation, payment recall attempts, legal decisions, customer protection, regulator engagement, and clear internal communication. These decisions must happen within hours, not weeks.
2) Enhance forensic competency and evidence capability
Ensure on-call forensic teams, whether inhouse or external consultants, are equally adept at analysing both paper and digital trails. Preservation and analyses of emails, devices, logs, cloud records, chat data, online evidence, payment trails, and cryptocurrency records must be quick and defensible.
3) Embrace anti-fraud by design principle
Proactively assess fraud risks and implement effective controls for new products, channels, payment flows, onboarding journeys, AI tools, and supplier processes robustly before launch, and eliminate the opportunity for criminals to exploit the weak points.
Ramesh Moosa, CA (Singapore), is ISCA’s Financial Forensic Accounting Oversight Committee (FFAOC) Member, Independent Advisor and Trainer.
