Skip to main content

Clari5

India Regional Intelligence Brief | October 2026 Roundup – Newsletter

India Regional Intelligence Brief | October 2026

For risk, compliance, and fraud leaders at Indian banks

Summary

Regulators and courts are setting a shared standard: act fast on a fraud report, show the evidence, and stop mule accounts early. The Supreme Court wants RBI’s new cyber-fraud rules to carry a liability clause, and RBI’s customer-protection directions start January 1, 2027. The timeline gives banks room to prepare.

Rules are taking shape, with time to prepare

    • Supreme Court (October 29): The Court asked the RBI to include a liability clause in its cyber-fraud rules, with compensation of at least the amount lost where a bank’s due diligence falls short, plus safeguards against unusual transactions. The Attorney General proposed notification by November 30 and operation by December 31. Source
    • RBI customer-protection directions (January 1, 2027, one-year scheme): For electronic transfers up to ₹50,000, victims who report within five days can receive 85% of net loss, capped at ₹25,000. The bank carries the burden of proving customer liability. On losses below ₹29,412, RBI funds 65% and the victim’s bank and the beneficiary bank pay 10% each, so early mule detection benefits every institution. Source
    • Draft mule-hold procedure: RBI’s September 11 draft sets time-bound debit holds on suspected mule transactions or accounts, for up to 60 days, effective April 1, 2027 if finalized as drafted. Final directions are pending, with December reported as a target. Source

Worth reviewing: Whether hold, alert, and complaint workflows can run to these timelines with clear audit trails.

Courts ask for evidence and a prompt response

Recent rulings share one theme: outcomes turn on whether the bank can show system records and a prompt response after a report.

  • Kerala High Court (October 5): ₹3.73 lakh refunded to a customer who shared a one-time password (OTP) under deception. The OTP alone does not prove negligence. Source
  • Kalaburagi consumer commission: ₹4.84 lakh refunded because records did not establish that an OTP was shared. Source
  • Barnala consumer commission: ₹3 lakh plus ₹20,000, with the commission looking for a record of steps to trace the beneficiary and recover funds. Source

Worth reviewing: Whether authentication, complaint handling, and recovery steps can be reconstructed quickly in a contested case.

Mule networks and insiders: what recent cases teach

  • Payment-message integrity (alleged): In a ₹13.96 crore case, 187 transfers on October 15 to 17 reached 107 accounts across 36 banks. Maharashtra Cyber alleges the Real Time Gross Settlement (RTGS) system was tampered with. Receiving banks are not named, so any institution may hold beneficiary accounts. Lesson: monitor payment-message integrity, not only customer behavior. Source
  • Insider and privileged-user risk: In a suspected ₹25 crore case, one employee allegedly used colleagues’ credentials for nearly three years and switched off customer alerts. It surfaced when a customer updated a passbook. Lesson: monitor shared credentials, staff overrides, and live Know Your Customer (KYC) checks at onboarding. Source
  • Linked accounts: In Assam, an account allegedly moved lakhs over 47 days with near-instant withdrawals and was identified through law-enforcement intelligence. In Champawat, 148 linked accounts across several banks were cashed out through ATMs and point-of-sale machines. Lesson: combine bank-side signals with external intelligence. Assam | Champawat

Worth reviewing: Inbound credits and onward movement for October 15 to 17, and links between accounts across institutions.

Supervisors are raising the bar on cyber and API assurance

  • India Today, citing Reuters and unnamed sources, reports that RBI wants banks to evidence cyber compliance, review application programming interface (API) links with NBFCs and fintechs, and detail readiness for attacks that use AI. No circular has been issued. Reported triggers include an advisory from the Indian Computer Emergency Response Team (CERT-In) of October 9. Source
  • Separately, the Economic Times reports about ₹300 crore in losses and about 150 complaints in two months from AI-enabled attacks on small NBFCs. Banks that host NBFC accounts are well placed to detect these flows in flight. Source

Worth reviewing: Whether payout and partner APIs receive the same monitoring as customer channels.

Global context: the same direction

  • EU Court of Justice (case C-70/25): A bank must refund an unauthorized payment immediately, and by the end of the first working day after it learns of it, provided the customer notifies without delay. Suspected gross negligence is not a ground to refuse, and the bank may seek recovery later. Source
  • UK Financial Conduct Authority: Firms closed 238,396 suspected mule accounts in 2025. Criminals usually cash out between the second and fifth account, and older accounts made up a larger share of closures at retail banks, so mule monitoring pays off on existing relationships too. Source
  • Financial Action Task Force (FATF) plenary: The October plenary approved a report on scam compounds, drawing on more than 100 case studies, to help countries disrupt the fraud supply chain. FATF’s President noted that in Thailand funds can leave the financial system within minutes, while victims often take about 20 hours to realize they have been scammed. He added that instant payments and AI-enabled impersonation make money easier to move and hide. FATF | ANI

Dates

  • Expected by about November 8: India AI regulation consultation paper (announced October 8, “within a month”).
  • November 30: Supreme Court hearing; RBI cyber-fraud rules proposed for notification.
  • December (target, unconfirmed): Final RBI mule-account directions; all banks on the Indian Cyber Crime Coordination Centre (I4C) portal.
  • December 31: Proposed date for RBI cyber-fraud rules to operate.
  • January 1, 2027: Customer-protection directions start.
  • April 1, 2027: Draft mule-hold procedure takes effect, if finalized as drafted.

Clari5 is glad to walk through any of these developments with your teams.

Africa Regional Intelligence Brief | October 2026 Roundup – Newsletter

Africa Regional Intelligence Brief | October Roundup

Summary

Four themes run through this month’s developments. Regulators are setting dates while scope is still being clarified. Losses are concentrating in mobile, wallet, and SIM channels. Banks are being asked to prove their controls and to stop mule and insider flows early. New payment rails are widening exposure. Nigeria’s data localization deadline is two months away, so scope questions are worth settling now.

1. Regulators are setting dates, and scope is still being clarified

  • Nigeria: The Central Bank of Nigeria (CBN) requires payment data generated in Nigeria to be stored locally from January 1, 2027, and confirms the date has not moved. In a regulator survey, 22 of 44 institutions had not begun, with unclear scope the main obstacle, and CBN plans to issue FAQs. Automated anti-money laundering (AML) systems integrated with customer identity data are due for banks by September 10, 2027. Source
  • Kenya: The draft National Payment System Bill would add licensing, trust accounts for wallet funds, and immediate reporting of material cyber incidents. Consultation closed October 9. Comments on draft revised risk management guidelines are due November 7. Source
  • Ghana: The Bank of Ghana plans thematic cyber compliance reviews and a directive on AI in financial services, with no timeline published.  Source

Worth reviewing: Scope assumptions against CBN’s FAQs once issued, and whether detection models can be documented and explained on request.

2. Losses are concentrating in mobile, wallet, and SIM channels

  • Uganda: 6,533 fraud and cyber incidents and UGX 24 billion in losses from January to August 2026, led by peer-to-peer and wallet fraud. Source
  • Kenya: INTERPOL reports SIM swap investigations up 327% in 2025, with weak identity checks at SIM issuance as the enabling gap. Source
  • Ghana: The Cyber Security Authority recorded GH¢209.7 million (about USD 18 million) in cybercrime losses from January to September. Source
  • Nigeria: A platform generating fake payment receipts, promoted to 61,000+ Telegram subscribers, exploits delays in SMS and bank alerts. Source
  • Enforcement progress: A joint operation by 15 technology companies and law enforcement from 10 countries, announced October 13, produced 28 arrests in Ghana and Nigeria and removed 3.7 million accounts. Source

Worth reviewing: Whether SIM change and device signals reach transaction decisions before funds move, and how a customer manipulated into authorizing a payment is detected.

3. Banks are being asked to prove their controls and stop mule flows early

  • Ghana: On October 16, the Ghana Association of Banks urged real-time fraud intelligence sharing, noting that in zero-recovery cases the next institution sometimes learned of the fraud two days later. Officials also stressed that outsourcing a service does not outsource accountability, and bank specialists want controls checked before customers are blamed. GNA, reviews | GNA, AI rules
  • Nigeria: The Nigerian Financial Intelligence Unit reports suspicious transaction reports down 49% in 2025 while cash transaction reports rose 62%, and flags accounts opened in one name but controlled by another. Source
  • Mule accounts: A South African retail bank holds funds in identified mule accounts while keeping them looking active, and notes that biometrics need to work alongside transaction monitoring. Source

Worth reviewing: Whether the controls behind a disputed transaction, including outsourced ones, can be evidenced quickly, and how mule signals are monitored.

4. New rails are widening exposure

  • Ethiopia: The National Bank joined the Pan-African Payment and Settlement System on October 15, enabling local-currency settlement without correspondent routing. Non-bank wallet providers need a letter of no objection before connecting. Source
  • Kenya: PesaLink ID aliases such as phone numbers and email addresses raise number recycling and misdirected payment risks. Source

Worth reviewing: How screening, monitoring, and investigation responsibilities are mapped across new corridors and aliases.

Global context

  • FATF: Kenya stayed on the grey list after the October plenary, with two items left, including more risk-aligned money laundering prosecutions. FATF
  • Shared intelligence: Canada launched a bank and mobile operator fraud exchange on October 14, and India, Singapore, and Malaysia are piloting shared mule intelligence among banks. Worth reviewing: whether your bank is in a shared-intelligence channel. 
  • Liability and lookbacks: The EU Court of Justice (C-70/25) held that suspected gross negligence does not justify refusing a refund of an unauthorized payment. A large US bank was penalized USD 1.21 billion for late suspicious activity reporting, with an independent lookback required. Worth reviewing: whether alert rules are tested against activity they missed, and what evidence supports a refund decision.

Dates to watch

Date Milestone
November 7, 2026 Kenya: comments due on draft risk management guidelines
End-November 2026 (expected) South Africa: payments authorization framework
December 31, 2026 Nigeria: beneficial ownership disclosure. Ethiopia: risk-based capital rules
January 1, 2027 Nigeria: data localization takes effect
September 10, 2027 Nigeria: automated anti-money laundering compliance due for banks

Clari5 is glad to walk through any of these developments with your teams.

This brief summarizes public reporting and should not be treated as legal or regulatory advice.

 

When A Customer Is Tricked Into Approving A Payment Who Pays?

When A Customer Is Tricked Into Approving A Payment Who Pays?

A CUSTOMER SHARED HER SCREEN WITH A FRAUDSTER. THE BANK WAS ORDERED TO REFUND HER

In 2024, a woman in Delhi received a call from a fraudster and allegedly shared her phone screen. The next morning, two transfers followed: ₹5 lakh and ₹44,488, authenticated with her PINs and OTPs. The bank argued that the payments were authenticated and the loss was hers. On October 28, a Delhi consumer commission disagreed. An OTP proves that the correct code was entered, not that the customer understood or intended the payment. The commission acknowledged that screen-sharing could amount to negligence, but found that the bank had not established that the entire loss was her responsibility. It also found that the bank had not produced sufficient device, IP address, beneficiary and audit records. The bank was ordered to refund ₹4,14,488, with interest and compensation. Authentication alone is not proof of consent. Banks need records to establish what happened and who was responsible.

FROM JANUARY 1, BANKS MUST ESTABLISH WHEN A CUSTOMER IS LIABLE

New directions from the RBI, issued in June, apply to digital payments made on or after January 1, 2027. They extend fraud protection to payments customers make after being deceived or coerced, not just transactions they never initiated. The burden of proving customer liability rests with the bank. Banks must examine complaints and determine liability within 45 days for domestic transactions and 60 days for cross-border transactions. This makes the bank’s investigation and transaction records central to deciding who bears the loss.

REPORT A THIRD PARTY BREACH WITHIN FIVE DAYS AND THE CUSTOMER HAS ZERO LIABILITY

Under the new rules, customers who report a third-party breach within five calendar days have zero liability, and the transaction must be reversed. A third-party breach is one where neither the customer nor the bank is at fault. After five days, liability is determined under the bank’s own policy. Other cases depend on where the fault lies. Banks must also send instant SMS alerts for digital transactions above ₹500. For credit card complaints, they must provide temporary credit within five days. Timely alerts matter: customers cannot report transactions they do not know have occurred.

THE SUPREME COURT WANTS EARLY HOLDS ON SUSPECT ACCOUNTS AND BANKS TO PAY WHEN THEY MISS THEM

At an October 29 hearing on “digital arrest” scams, where fraudsters impersonate police officers and pressure victims into transferring money, the Supreme Court pressed the RBI to finalize its draft KYC rules. The Attorney General told the Court the rules would be notified by November 30 and take effect by December 31. The draft would allow banks to temporarily hold outgoing payments from suspected mule accounts (accounts used to receive and pass on stolen money) for up to 60 days unless police direct otherwise. Account holders would have 20 days to explain the transactions. The Court also said banks should compensate victims, at least for the amount lost, where failures in due diligence enabled the fraud. It noted that bank employees were involved in some scams. The Court is yet to issue further orders. A temporary hold can prevent further loss, but only if the bank identifies the suspect account in time.

CRIMINALS USED LENDER’S OWN CREDENTIALS TO MOVE MONEY. MANY CHECKS WERE BYPASSED.

Not all fraud involves a customer. On October 9, India’s national cyber agency, CERT-In, warned of attacks on lenders, fintechs and wallet operators. Criminals stole the credentials these firms use to move money, then initiated transfers through their own pre-approved systems. Because the requests came from trusted systems, they bypassed many application-level checks. CERT-In noted the likely use of AI-assisted tools and warned that the time between breaking in and stealing funds is shrinking. Separately, a media report on October 29, estimated that smaller non-bank lenders had lost about ₹300 crore to cyberattacks in recent months. With no customer to blame, the loss fell on the lenders themselves. A valid credential establishes access, not whether a transaction is legitimate.

The Account Belongs To One Person But Someone Else Is Using It

The account belongs to one person. Someone else is using it.

 4,086 ACCOUNTS, ONE QUESTION: WHO IS REALLY USING THEM?

Madhya Pradesh cyber police have launched a statewide drive, running until November 28, targeting 4,086 suspected first-layer mule accounts. These are the first accounts to receive stolen money before it is moved further through the network. For the first time, the drive focuses on cases where both the victim and account holder are from the state. In the earlier round, account holders from other states made arrests and prosecution more difficult. Investigators will now examine whether these accounts were originally legitimate or opened for fraud, whether genuine documents were used, who persuaded the holders to hand over account access for a commission, and where the money went next. The previous drive resulted in 26 cases across 12 districts, 46 arrests and ₹13 lakh in frozen funds. The first receiving account is often the bank’s earliest opportunity to stop the money from entering the wider mule network.

83 ACCOUNTS ACROSS 22 BRANCHES, ALLEGEDLY SOLD FOR ₹20,000 TO ₹25000 EACH 

Hyderabad police have busted a gang that allegedly bought bank accounts from local residents and resold them to cyber fraudsters. Each account kit reportedly included a passbook, ATM card, PIN, account details and a SIM card linked to the account, all sent onwards by courier. Police identified 83 such accounts across 22 bank branches. Eleven are already linked to 14 reported cyber fraud transactions involving about ₹10 crore across several states. The worrying part is how ordinary these accounts may have looked individually. The connections between account holders, linked SIMs, devices and subsequent transactions are what can expose the larger network.

A ₹10 LAKH SCAM LED TO 255 COMPLAINTS AND ₹72 CRORE

Mumbai police began with a ₹10 lakh impersonation scam. Following the trail through bank accounts, phones and cryptocurrency transactions led them to a network linked to 255 complaints across the country and approximately ₹72 crore. Six people from three states have been arrested. Police allege that members traded bank accounts and SIM cards in messaging groups and persuaded account holders to install malicious app files that allowed others to operate their accounts. Police have recovered ₹5.5 lakh so far and are still establishing how individual accounts were controlled. The accounts were genuine. The real question is who had control of them, and how quickly the money was moved once it arrived.

EIGHT BANK EMPLOYEES, DIFFERENT BRANCHES, ONE ALLEGED PURPOSE

West Bengal police have arrested a bank deputy manager and seven other employees for allegedly opening mule accounts and routing fraud proceeds in return for commissions. The eight employees were attached to different branches, making the case particularly difficult to spot through branch-level reviews alone. This is not the first such case. In April, Hyderabad police arrested 52 people in a mule account network, 32 of them bank employees. The Supreme Court raised the same concern on October 29. Mule account investigations cannot stop at the account holder or the transaction. They may need to look inside the bank too, connecting account-opening activity, employee involvement and common beneficiaries across branches.

SINGAPORE TOUGHENS SENTENCES FOR YOUNG MULES. BANKS STILL HAVE TO SPOT THE ACCOUNTS.

On November 3, Singapore introduced a bill that would make imprisonment or reformative training the default sentence for offenders aged 21 or younger convicted of scam-related or scam-enabling offences, unless exceptional circumstances apply. It is not yet law. The number of youths investigated for cheating and related offences rose 37%, from 814 in 2024 to 1,114 in 2025. Most were reportedly mules who handed over bank accounts, national digital ID logins or SIM cards. Singapore police receive about 90 scam reports and record about S$2 million in losses every day. Tougher sentences may deter young people from becoming mules, but they do not remove the underlying vulnerability: an account handed over to someone else can become an active part of a fraud network long before its legitimate owner realises what is happening.

Press Release – Mashreq launches Generative AI capabilities to strengthen fraud risk management with Clari5 – Duplicate – [#75253]

Press Release – Mashreq launches Generative AI capabilities to strengthen fraud risk management with Clari5

Dubai, UAE, 03 November 2026: Mashreq, a leading financial institution in the MENA region, has launched new generative AI capabilities within its fraud platform in collaboration with Clari5 – A Perfios Company. The move brings tangible benefits across three priority areas: enhancing fraud detection effectiveness, improving analyst productivity, and supporting stronger investigative decision-making.

Announced at Mashreq’s Innovation Hub in Dubai in the presence of senior leadership from both organizations, the launch demonstrates how generative AI can be applied within a banking environment to complement the expertise and judgement of fraud professionals.

Ahmed Abdelaal, Group Chief Executive Officer at Mashreq, said: “In an increasingly digital financial landscape, fraud is becoming more sophisticated, and the way banks respond must adapt accordingly. At Mashreq, we are committed to implementing AI responsibly in areas where it can have a meaningful impact, strengthening the resilience of our operations and the trust our customers place in us.”

Nitin Chugh, MD and Group CEO, Perfios, said: “Our collaboration with Mashreq demonstrates how generative AI can help banks tackle the growing complexity of financial crime by bringing together disparate information, accelerating investigations and enabling better-informed decisions. The goal is to augment investigators’ expertise, not automate judgement, while ensuring AI operates with the rigour and accountability that financial crime management demands.”

Rivi Varghese, Co-Founder and CEO, Clari5, said: “Banks don’t need AI that acts on its own. They need AI they can govern by their own rules. Clari5 has shown how AI can operate on-premises inside a bank’s own environment, grounded in the bank’s context, and remain auditable, explainable and under human control. We are proud to have gone live with Mashreq and set an example of how enterprise-grade AI should operate in managing financial crime.”

Rania Nerhal, Group Chief Client Experience & Conduct Officer at Mashreq, said: “The value of generative AI lies in its ability to help us interpret relevant information and respond to emerging risks with greater speed, consistency, efficiency, and precision. This launch is part of our continued investment in deploying the latest technology to evolve our fraud risk management framework.”

Developed through the ongoing collaboration between Mashreq and Clari5, the new capabilities further strengthen the bank’s fraud detection platform. They are designed to deliver two key benefits:

  • Seamless customer experience: Supporting faster, more consistent, and more effective fraud detection.
  • Greater consumer trust: Reinforcing confidence in the bank’s ability to protect customers and respond to emerging risks.

About Mashreq:

Mashreq is a leading financial institution headquartered in Dubai, serving individuals, entrepreneurs, corporates, and institutions across the UAE and key regional and international markets. With a heritage spanning nearly six decades, Mashreq combines digital-first innovation with deep relationship banking across retail banking, corporate and investment banking, global markets, transaction banking, Islamic banking, wealth management, and treasury. The bank delivers secure, advanced, and customer-centric experiences while actively driving the transition toward a sustainable economy. Through a robust Environmental, Social, and Governance (ESG) framework, Mashreq is committed to facilitating sustainable finance, fostering responsible operations, and delivering positive societal impact.

With an extensive international footprint spanning the UAE, Oman, Qatar, Kuwait, Bahrain, Egypt, Pakistan, Türkiye, India, the UK, the US, and Hong Kong, Mashreq seamlessly connects regional businesses with global markets while maintaining deep local expertise.

Guided by its core purpose to help customers Rise Every Day, Mashreq partners with clients to navigate change, unlock sustainable growth opportunities, and achieve long-term success.

For more information, please visit: www.mashreq.com For media inquiries, please contact: media@mashreq.com

About Clari5:

Clari5 – A Perfios Company, is the leading provider of real-time financial crime management platforms for banks, insurance companies, telcos and other financial institutions worldwide. Powered by AI-native capabilities, Clari5 helps detect, prevent, report and investigate financial crime in real time. Trusted by 80+ banks across 30+ countries for over two decades, Clari5 unifies fraud and AML on a single platform. It processes 12+ billion transactions every month and protects more than 1 billion accounts. Clari5 is part of the Perfios group, India’s largest B2B SaaS TechFin, serving over 1,000 financial institutions across 18 countries. To know more, visit www.clari5.com.

Mashreq launches Generative AI capabilities to strengthen fraud risk management with Clari5

Dubai, UAE: Mashreq, a leading financial institution in the MENA region, has launched new generative AI capabilities within its fraud platform in collaboration with Clari5 – A Perfios Company. The move brings tangible benefits across three priority areas: enhancing fraud detection effectiveness, improving analyst productivity, and supporting stronger investigative decision-making.

Announced at Mashreq’s Innovation Hub in Dubai in the presence of senior leadership from both organizations, the launch demonstrates how generative AI can be applied within a banking environment to complement the expertise and judgement of fraud professionals.

Read more

BNM RMiT Appendix 11

BNM RMiT Appendix 11: What Malaysian Banks Must Do to Detect and Stop Digital Fraud

Short answer: Appendix 11 of Bank Negara Malaysia’s (BNM) Risk Management in Technology (RMiT) policy document sets minimum fraud detection standards for retail digital banking. Banks must build a risk profile for each customer, detect and block suspicious transactions in real time, contact the affected customer within 30 minutes of a block, and update fraud rules promptly as new techniques appear.

What is Appendix 11?

Appendix 11 is the fraud detection section of the RMiT policy document (reference BNM/RH/PD 028-98), issued on September 25, 2026. The document supersedes several earlier instruments, including the Specifications on Fraud Detection Standard dated March 27, 2024. Fraud detection now sits inside the main technology risk framework, and paragraph 12.4 links the two. It requires fraud detection capabilities and rules to be updated in a timely manner when new modus operandi appear.

Who does it apply to?

Appendix 11 covers financial transactions carried out through digital services for retail customers. BNM’s RMiT FAQ (updated July 1, 2026) adds that for corporate and other non-retail segments, the controls apply on a risk-based and proportionate basis.

What does Appendix 11 require?

Area

What BNM requires

What it means in practice

Customer risk profile

A detailed profile for each customer, using demographics, geography, transaction history and behavior such as typing speed and mouse movement

Each transaction is judged against that customer’s own baseline

Real-time action

Detect and block suspicious transactions in real time, based on individual profiles

The decision happens before the money moves

Event triggers

Monitor new e-banking enrollment, new device registration, password reset, limit increase, contact detail changes and new payees

A risky event raises scrutiny of what follows

Transaction patterns

Watch for high-volume debits to new beneficiaries, sudden pattern changes, activity on dormant accounts, unusual locations and payments to suspected mule accounts

Behavior over time matters as much as any single payment

Device and identity

Detect device fingerprint changes and attempts to bypass fingerprinting, account takeover using advanced artificial intelligence tools, changes to biometric credentials, and score vulnerable customers higher

Device and identity signals feed the same risk decision

Sessions and consent

Terminate hijacked sessions, warn customers about risky applications, and obtain consent before device profiling

Digital trust extends beyond the payment

Investigation and response

Investigate by priority, verify with the customer before releasing a flagged transaction, notify immediately on a block, and make contact within 30 minutes, including peak periods

Operations must keep pace with detection

Continuous updates

Enhance rules promptly when new techniques evade detection, and review thresholds against new typologies, including those seen in other jurisdictions

Fraud controls change as fast as fraud does

Playbook

A fraud management playbook, validated at least once a year using confirmed and missed fraud

Every case feeds the next improvement

What happens if a bank has gaps?

Paragraph 12.2 sets out the consequence. A bank that has not implemented the specified digital services controls must be ready to show that its alternatives are equivalent or superior, and it must assume liability for fraud arising from the gap. Appendix 11 also encourages banks to exceed the minimum standards and keep upgrading their detection capability, so the requirements are a baseline, not a target.

Five questions to test readiness

  1. Can the bank score every digital transaction against that customer’s own profile before funds move?
  2. Does the system treat a new device, password reset, limit increase or new payee as a signal, and track what happens next?
  3. Can it detect device changes and attempts to bypass device fingerprinting?
  4. Can operations verify and contact the customer within 30 minutes, including at peak hours?
  5. How long does a new fraud rule take to move from a fraud team finding to production, and when was the playbook last validated?

How Clari5 supports Appendix 11

Clari5 provides real-time enterprise fraud risk management to 80+ banks across 30+ countries, and Chartis Research named Clari5 a Category Leader in Enterprise Fraud Solutions in its 2026 Quadrant Update.

Clari5 Enterprise Fraud Risk Management (EFRM) addresses the main Appendix 11 requirements as follows.

Appendix 11 requirement

Clari5 EFRM capability

A risk profile for each customer

Builds and maintains an individual profile from transaction history, timing, velocity, new beneficiaries and geography, and captures behavioral signals

Real-time detection and blocking

Blocks a digital transaction in the payment path, not only raises an alert

Event and pattern triggers

Treats events such as new device registration, password reset, limit increase and new payee as risk signals, and tracks the transactions that follow

Device and session risk

Uses device fingerprinting & behavioral biometrics to provide device and session signals, detect device fingerprint changes and attempts to bypass fingerprinting, and support termination of hijacked sessions

Account takeover and vulnerable customers

Detects account takeover and identity fraud and applies higher risk scores to vulnerable customer segments

Mule and external intelligence

Flags transactions to suspected mule accounts, where the bank holds the relevant account details from its own, industry or law enforcement sources

Investigation and customer contact

Case management prioritizes alerts by risk level and supports call-back verification

Rule changes and playbook

Lets fraud teams add and tune rules promptly, and feeds confirmed and missed fraud outcomes back into thresholds and the annual playbook review

Frequently asked questions

When was the updated RMiT policy document issued?

BNM issued the updated RMiT policy document (reference BNM/RH/PD 028-98) on September 25, 2026. It updates the revision BNM issued in November 2025. The effective date remains November 28, 2025, unless a provision states a different date.

How fast must banks contact customers after a blocked transaction?

Within 30 minutes of the transaction, including during peak periods (Appendix 11, paragraph 5). The bank must also notify the customer immediately when it blocks a transaction.

Does BNM require banks to use artificial intelligence for fraud detection?

No. Appendix 11 requires banks to detect account takeover and identity fraud carried out with advanced artificial intelligence tools. It does not prescribe a specific technology for detection.

Does Appendix 11 apply to corporate banking?

It is aimed at retail digital services. BNM’s FAQ says non-retail segments apply the controls on a risk-based and proportionate basis.

Why Egyptian banks and insurers need to connect Fraud, AML, Cybersecurity and AI

The criminal sees one attack. Your institution may see five alerts.

It may start with a stolen mobile credential. Then comes a new device, an unusual login, a first-time beneficiary and an instant transfer. The money may move through several accounts before the first customer complaint is raised.

The Cybersecurity team may see the compromised device. Fraud may see the unusual payment. AML may see a possible mule account. Customer service may see the complaint.

The criminal sees one connected attack. The institution sees separate alerts.

That gap is becoming one of the most important financial crime challenges for Egyptian banks and insurance companies. Fraud, money laundering and financial cybercrime can be connected stages of the same event. The institutions that connect the signals early will be better placed to protect customers, stop the movement of funds and meet supervisory expectations.

This is the next phase of the journey. The question is no longer only how to build a stronger fraud function. It is how Fraud, AML, Cybersecurity, Compliance and Customer Protection can share the right intelligence quickly enough to act as one line of defence, while keeping their separate responsibilities clear.

Egypt’s digital shift is shrinking the response window

Egypt’s financial system is rapidly evolving and getting digitized. CBE data shows financial inclusion reached 79% in June 2026, meaning 56.4 million citizens had active financial accounts. The same direction is visible in digital payments, where instant payments, mobile wallets and contactless payments are operating at significant scale.

By June 2025, CBE-reported figures showed more than 16 million InstaPay users, more than 1.1 billion transactions worth EGP 2.4 trillion, and 55.5 million electronic wallets processing 1.4 billion transactions worth EGP 1.8 trillion. CBE digital transformation figures reported from PAFIX 2025

The CBE’s 25 June 2026 circular on specialised anti-fraud units is another signal that fraud capability remains a material supervisory priority. CBE circular on specialised anti-fraud units

The implication for financial crime leaders is practical: the faster the payment, onboarding and recovery journey becomes, the less time there is for a control that depends on end-of-day review or manual hand-offs. A compromised account can add a beneficiary, initiate a payment and become part of a wider network before separate teams have assembled the full picture.

The CBE’s 25 June 2026 circular, following its earlier circular on establishing specialised units in banks for managing and combating fraud, is another signal that fraud capability remains a material supervisory priority. The next operating challenge is to make that capability work with the other teams that see different parts of the same risk.

The gap is between functions, not inside one function

The answer is not to remove the distinction between Fraud, AML and Cybersecurity. Each function has its own mandate, expertise and accountability. The opportunity is to connect their intelligence, decision points and escalation workflows.

The operating principle should be simple: separate accountability, shared intelligence.

For an AML/CFT function, statutory and regulatory responsibilities remain intact. Fraud and Cybersecurity teams contribute the signals and context that can help AML investigators understand the wider pattern. The same principle works in the other direction. AML alerts, confirmed cases and network relationships can improve fraud and cyber decisions.

This is especially relevant as Egypt’s regulatory environment continues to evolve. MENAFATF’s Fourth Enhanced Follow-up Report for Egypt was published in May 2025, providing the latest follow-up assessment used here. It should be treated as an authoritative assessment of Egypt’s AML/CFT progress, not as a substitute for 2026 regulatory updates.

What a shared risk view should connect

  • Customer and account identity.
  • Device, session and network behaviour.
  • Transaction, payment and beneficiary history.
  • Agent, merchant, employee and intermediary activity.
  • Claims, policies and payment accounts for insurance.
  • Previous alerts, investigations and confirmed decisions.
  • Cyber incidents and confirmed account compromise.

A single signal rarely proves fraud. A combination of signals can change the decision.

Consider a customer who logs in from a new device, changes a mobile number, adds a new beneficiary and sends an instant transfer outside the normal pattern. A transaction-only control may see one unusual payment. A connected risk view can identify a possible account-takeover chain and bring the right teams into the decision earlier.

The same logic applies to insurance. A claim should not be assessed only on the claim itself. The relationship between the claimant, policy, agent, device, bank account, service provider and previous claims can reveal patterns that a single claim-control process cannot see.

Cybersecurity is part of the financial crime picture

A cyber event does not always end when access is restored. A compromised credential can become the starting point for fraud, account takeover, mule activity or suspicious movement of funds.

For insurers and other non-bank financial institutions, this connection is also becoming more important. FRA Resolution No. 227 of 2025 requires relevant non-banking financial entities to strengthen technology infrastructure and cybersecurity capabilities, including periodic penetration testing and annual information-security reporting. The resolution reinforces the need to treat cyber resilience and financial risk as connected management issues.

The goal is not to create one large team. It is to make sure the teams do not lose the story between them.

How AI can help detect and prevent fraud

Fraud is moving faster than manual controls can respond.

A suspicious transaction rarely appears alone. It may follow a new device, unusual login, changed customer details or a new beneficiary. The challenge is not generating more alerts. It is recognising which signals belong to the same risk.

This is where AI can make a practical difference.

AI can analyse customer behaviour, transaction patterns, device activity and account relationships to identify unusual activity, connect related signals and prioritise cases that need attention. It can also learn from previous investigations to help teams recognise emerging fraud patterns and reduce repetitive investigation work.

For banks and insurers, the value is simple: detect earlier, investigate faster and intervene before losses escalate.

But AI is not a replacement for strong financial crime controls. If Fraud, AML and Cybersecurity data remain fragmented, AI may simply process fragmented information faster.

The stronger model is connected intelligence first, AI second, human judgement throughout.

AI should strengthen judgement, not replace it

AI can help financial crime teams identify unusual behaviour, connect related entities, prioritise investigations and reduce repetitive analyst work. Its value is greatest when volumes grow faster than human investigation capacity.

But AI is not a shortcut. If the underlying signals remain fragmented, AI can simply make a fragmented institution faster.

The stronger model is: connected data first, AI second, human judgement throughout.

Senior leaders should ask:

  • Can an investigator understand why a case was prioritised?
  • Can the institution explain important decisions to internal audit and supervisors when required?
  • Are models tested for data quality, false positives, drift and unintended outcomes?
  • Is a named human owner accountable for high-impact decisions?
  • Can model outputs be linked back to the evidence used in the decision?

AI should identify and prioritise. Experienced people should investigate, decide and act.

The next three years: Connect → Predict → Orchestrate

Phase Leadership Priority
2026 | CONNECT Map the full attack journey. Connect identity, device, transaction, beneficiary, cyber, AML and investigation signals. Establish common risk language and clear ownership.
2027 | PREDICT Use behavioural intelligence, network relationships, historical investigations and governed AI to identify emerging patterns earlier and prioritise the cases that matter most.
2028+ | ORCHESTRATE Move from isolated alerts to coordinated intervention: signal → risk decision → action → investigation → learning. Measure the complete outcome, not only the number of alerts.

This does not mean replacing every existing system. It means first asking whether the current environment can share the data, risk signals, cases and decisions required to act at the speed of the risk. Integration or consolidation should follow that assessment.

Measure what the customer and the business actually experience

Alert volumes can grow without improving protection. A more useful leadership dashboard should track:

  • Time to detect
  • Time to intervene
  • Prevented loss
  • Recovery rate
  • False-positive rate
  • Investigation productivity
  • Escalation quality
  • Customer friction

These measures connect financial crime performance to customer protection, operational efficiency and board-level outcomes.

The competitive advantage will be trust at transaction speed

Over the next few years, banks and insurers will compete on trust as much as on products. Customers will remember whether an institution protected them quickly, communicated clearly and treated legitimate transactions fairly.

The strongest financial crime operating models will therefore be real-time across relevant channels, connected across Fraud, AML and Cybersecurity, explainable to investigators and supervisors, adaptable to new attack patterns and measurable at leadership level.

The strategic question is no longer simply, “Does the institution have a fraud system?”

It is: “How quickly can the institution understand and stop a connected financial crime attack while protecting legitimate customers?”

Egyptian financial institutions do not need another disconnected control. They need a clear operating model in which specialist teams can see the same risk early enough to act.

The next competitive advantage may not be who generates the most alerts. It may be who understands the customer’s risk fastest, intervenes responsibly and protects the legitimate customer without unnecessary friction.

The criminal already sees the chain. The opportunity for financial crime leaders is to make sure their institution does too.

Frequently Asked Questions (FAQs)

What is fraud and AML convergence?

It is the coordinated use of fraud, AML, cyber, identity and transaction intelligence to identify connected financial crime activity. Teams keep their separate responsibilities but work from shared risk information.

How can Egyptian banks connect fraud, AML and cybersecurity?

Start by mapping the attack journey and identifying where identity, device, transaction, beneficiary, cyber and investigation information is separated. Then connect the signals, cases and escalation workflows needed for timely decisions.

Why is real-time fraud detection important in Egypt?

The rapid adoption of instant payments, mobile wallets and digital banking reduces the time available for manual review. Real-time monitoring can help institutions intervene before funds move through additional accounts.

How can AI support fraud and AML teams?

AI can help identify unusual behaviour, connect entities, prioritise investigations and reduce repetitive work. It should operate with clear governance, explainability, testing and human oversight.

Should banks replace their existing fraud and AML systems?

Not necessarily. The first question is whether existing systems can exchange the data, risk signals, cases and decisions required for timely action. Integration or consolidation should follow the operating-model assessment.

Does this approach apply to insurance companies?

Yes. Insurers can connect policy, claimant, agent, identity, device, payment and claims information to identify fraud, collusion and suspicious activity earlier.

Nigeria’s Next Fraud Defence: Bringing Fraud, AML and AI Together to Protect Customers in Real Time

Nigeria’s digital payment fraud losses fell sharply in 2025. But the next fraud challenge will not be solved by more alerts, more rules or more disconnected systems.

 

According to NIBSS data presented at the 2026 Nigeria Electronic Fraud Forum, losses from electronic payment fraud fell by 51% in 2025 to ₦25.85 billion, down from ₦52.26 billion in 2024. That is welcome progress. Yet social engineering remained a major threat, while internet banking recorded ₦13.37 billion in losses from 4,507 cases.

The lesson is clear: Nigerian banks are improving their defences, but criminals are changing how they operate.

A fraudster may begin with social engineering, compromise a customer’s device, move money through several accounts and finally cash out through a digital wallet or agent network. Fraud teams may see the suspicious transaction. AML teams may see unusual movement of funds. Customer service may receive the complaint. Investigators may hold the history of a similar case.

But if these signals remain in siloes, the bank just sees disconnected parts of the whole picture.

The criminal sees the whole picture.

That is why the next phase of banking financial crime protection must bring real-time Fraud prevention, AML and AI together on one connected platform, with the ability to detect risk and act while the transaction is still in progress.

The risk is no longer sitting in one channel

Nigerian banks already operate important controls:

  • Know Your Customer and customer due diligence.
  • Sanctions screening.
  • Transaction monitoring.
  • Fraud detection.
  • Customer risk scoring.
  • Case management.
  • Regulatory reporting.

The challenge is not the absence of controls. It is the gap between them.

A customer may have a valid identity and a long banking history. The device may appear familiar. The transaction may not breach a single rule. Yet the customer may suddenly receive funds from unrelated accounts, add a new beneficiary and transfer money to an account already associated with suspicious activity.

Viewed separately, each event may appear manageable. Viewed together, it may indicate account takeover, social engineering, mule activity or an organised fraud network.

This is where Fraud and AML leaders need a shared risk picture.

Customer risk should influence transaction risk. Transaction behaviour should update customer risk. Real-time Fraud intelligence should support AML investigations. AML intelligence should help Fraud teams identify connected activity. Investigation outcomes should improve future detection and prevention.

The objective is simple: One customer. One risk picture. One continuous view of financial crime.

What AI should mean in practice

For banking leaders, AI should solve practical problems that affect customers, investigators and regulators. Used responsibly, it can help banks:

  • Connect customer, account, device, beneficiary and transaction information.
  • Identify unusual relationships across accounts and channels in real time.
  • Detect changes in behaviour before losses occur.
  • Prioritise the cases that deserve immediate attention.
  • Explain why a transaction or customer relationship appears risky.
  • Find links between current alerts and previous investigations.
  • Summarise complex cases for faster review.
  • Support consistent documentation and regulatory reporting.

Early evidence from Nigeria supports this direction. Academic studies on AI-driven real-time fraud detection in Nigerian banks report that machine-learning based systems can identify anomalous transactions faster and with greater accuracy than traditional rule-based approaches, especially when combined with human oversight. [Nigerian AI fraud study]

The purpose is not to remove the investigator from the process. It is to remove the unnecessary searching, duplication and manual work that prevent investigators from applying their judgement where it matters most.

 

Regulation is moving towards timely, connected controls

The regulatory direction in Nigeria is becoming clear. The CBN has continued to strengthen expectations around AML, sanctions compliance, payment security, agent banking, customer protection and electronic fraud response. Its reforms also highlight the importance of safer payment systems and stronger customer protection.

Recent CBN initiatives include enhanced sanctions-related obligations, tighter oversight of payment channels and measures designed to improve the traceability and resilience of digital transactions. The regulator has also emphasised faster fraud response, with a reported expectation that banks should reduce response times to less than 30 minutes.

Nigeria has also strengthened its broader financial-crime framework. In October 2025, the Financial Action Task Force (FATF) removed Nigeria from increased monitoring after the country completed its agreed action plan on AML and counter-terrorist financing. [FATF announcement].

At the same time, reporting obligations are becoming more active. In 2025, Nigerian banks and fintechs filed 42,082 suspicious transaction reports with the Nigerian Financial Intelligence Unit (NFIU), with banks accounting for about 92% of all filings. [NFIU 2025 data]

This matters because a suspicious transaction identified after completion is not the same as one identified during authentication or payment processing.

For a bank, timely detection can mean:

  • Stopping or slowing a suspicious payment.
  • Protecting a customer from a social-engineering scam.
  • Preventing funds from moving through connected mule accounts.
  • Preserving evidence for investigation.
  • Improving the quality and timeliness of regulatory reporting.

Compliance and customer protection should not be treated as competing priorities. In a well-designed operating model, they reinforce each other.

Five priorities for Nigerian banks

1. Map the complete customer journey

Review risk from onboarding through authentication, account activity, beneficiary management, payments, complaints and investigation.

Identify where data is lost between Fraud, AML, Operations, Customer Service and Technology.

2. Connect the signals that matter

Prioritise the data that changes a decision:

  • Identity and KYC information.
  • Device and session behaviour.
  • Transaction patterns.
  • Beneficiary relationships.
  • Account linkages.
  • Customer complaints.
  • Previous fraud and AML investigations.

The aim is not to collect everything. It is to connect the information that helps the bank act.

3. Move from detection to intervention

Real-time detection has limited value if the bank cannot respond in real time.

Define clear actions for different levels of risk, including step-up authentication, transaction holds, customer confirmation, account restrictions, investigation escalation and regulatory reporting.

4. Use AI where human judgement is strongest

Begin with areas where investigators lose the most time:

  • Alert prioritisation.
  • Case summarisation.
  • Relationship discovery.
  • Similar-case identification.
  • Evidence gathering.
  • Quality assurance.

Keep human oversight for important decisions, especially where customers may be inconvenienced or accounts may be restricted.

5. Measure outcomes, not activity

Alert volumes alone do not show whether a financial-crime programme is effective.

Senior leaders should track:

  • Time taken to detect and respond.
  • Confirmed fraud prevented.
  • Investigation turnaround time.
  • False-positive rates.
  • Customer impact.
  • Connected cases discovered.
  • Quality of suspicious-activity reporting.
  • Explainability of important decisions.

The strongest banks will not necessarily be those generating the most alerts. They will be those making the best decisions with the signals they already have.

The next three years: from separate systems to shared intelligence

The direction of travel is straightforward.

Today: Fraud and AML teams often investigate separate alerts.
Next: They share intelligence across customers, accounts, transactions and cases.
Then: The bank assesses identity, behaviour, device, transaction and network risk together.

The goal: Detect risk, decide quickly, act in real time and explain the decision clearly.

This is also where competitive advantage will emerge.

A bank with strong Fraud controls but limited AML context still has a blind spot. A bank with effective AML monitoring but weak real-time transaction intelligence has another. A bank with both capabilities but disconnected investigation workflows may still respond too slowly.

The banks that stand out will be those that make protection visible in the customer experience:

  • Fewer legitimate transactions interrupted.
  • Faster support when customers report fraud.
  • Quicker recovery and containment.
  • More consistent decisions across channels.
  • Greater confidence from regulators, boards and customers.

That is how financial-crime defence becomes more than a compliance function. It becomes part of the bank’s reputation.

The leadership question

The question to consider is:

“Can we bring real-time Fraud, AML and AI together to see the complete risk picture before the customer pays the price?”

Criminals are already connecting identities, accounts, devices, beneficiaries and people. Nigerian banks must now connect their own intelligence faster. The future belongs to the banking institutions that combine strong controls with good judgement, timely action and a clear understanding of the customer behind every transaction.

Better intelligence. Better decisions. Faster protection. That is the standard Nigerian banks should build towards.

Frequently Asked Questions

What is Fraud and AML integration in banking?

Fraud and AML integration connects real-time fraud detection, transaction monitoring, customer risk, sanctions screening and investigation intelligence. It helps teams identify related activity as one financial-crime event rather than as separate alerts.

How can AI help Nigerian banks prevent fraud?

AI can help identify unusual behaviour, connect accounts and transactions, prioritise alerts and provide investigators with relevant context. It should support established controls and human judgement, not replace them.

Why is real-time fraud detection important?

Digital payments can move money within seconds. Real-time detection allows a bank to assess risk and intervene while a transaction, login or beneficiary change is still taking place.

What should banks do about existing Fraud and AML systems?

Banks do not necessarily need to replace every existing system. They can begin by connecting critical data, intelligence and workflows, then modernise progressively according to risk, business priorities and regulatory expectations.

What should Fraud and AML leaders measure?

They should measure detection quality, response time, investigation time, false positives, customer impact, connected-case discovery and decision explainability, rather than alert volumes alone.

Mule Accounts in Indonesia: Why Banks Detect Them Too Late

The account passed KYC. Then it moved Rp2 billion in 47 minutes.

This is not a hypothetical scenario. It is what happened at a major Indonesian bank in early 2026. The customer had valid e-KYC. The account sat dormant for 98 days. Then, in less than an hour, funds arrived from three different sources and were dispersed through five channels: BI-FAST transfer, QRIS payment, e-wallet top-up, card transaction and cash withdrawal.

By the time the fraud team received an alert, the money was gone.

This is the dormancy-to-activation gap, and it is the single biggest blind spot in Indonesian banks’ fraud and AML controls today.

This brief is for: Chief Compliance Officers, Heads of Fraud Risk, AML leaders, Chief Risk Officers and Digital Banking executives at Indonesian banks who need to close the gap between onboarding controls and real-time fraud detection.

The numbers that should worry every bank CXO

Indonesia’s scam response system shows both the scale of the problem and why speed matters.

From 22 November 2024 to 28 December 2025, the Indonesia Anti-Scam Centre (IASC) received 411,055 reports involving reported losses of Rp9 trillion. Around Rp402.5 billion was blocked or saved. CNBC Indonesia

By January 2026, OJK reported 432,637 complaints, reported losses of Rp9.1 trillion and Rp436.88 billion blocked. OJK also confirmed that Rp161 billion had been returned to 1,070 victims through funds blocked across 14 banks. CNBC Indonesia – Fund Recovery

In June 2026, media reports cited IASC figures of 608,168 reported accounts and Rp674 billion frozen, with approximately Rp200 billion returned to victims. Jakartaglobe

These are not just consumer protection statistics. They show how quickly fraud proceeds move through Indonesian bank accounts, virtual accounts, e-wallets and payment channels.

The central question for bank leaders: Can your bank recognise a risky change in account behaviour before the money leaves?

Why mule accounts pass your onboarding checks

A compliance head at a top-10 Indonesian bank put it plainly at a recent industry roundtable:

“Onboarding checks verify who a customer is, but not what the account will ultimately be used for.” Regulation Asia

This is the core problem. Mule networks in Indonesia are not relying on forged identities or fake documents. They are exploiting the gap between:

  • Day 1 verification (KYC, CDD, initial risk rating)
  • Day 90+ behaviour (dormancy, activation, cross-channel movement)

A mule account can sit inside your system for 90 to 120 days with:

  • A validated identity
  • A clean compliance record
  • No transaction red flags (because there are no transactions)

Then, almost overnight, it becomes a conduit for layered fund movement and rapid outflows.

The four-stage mule lifecycle in Indonesia

Mule networks follow a predictable pattern. Understanding this pattern is the first step to detecting it earlier.

Stage 1: Recruitment (Weeks 1-2)

The account holder is recruited, often through social media, informal intermediaries or false employment opportunities. They open the account believing it serves a legitimate purpose such as side income or delivery work.

Stage 2: Dormancy (Months 1-4)

Nothing happens. No transactions. No alerts. Nothing for a monitoring system to catch, simply because there is nothing to catch yet.

This is where most banks’ controls fail. Dormancy is treated as low risk by default. But for mule networks, dormancy is a feature, not a bug.

Stage 3: Activation (Hours 0-48)

Funds arrive from a compromised business account or an investment scam victim. Within 24 to 48 hours, the mule moves money across channels:

  • Core banking to e-money wallet
  • Wallet to card
  • Card to wire transfer
  • Multiple BI-FAST transfers
  • QRIS payments

This fragmentation is deliberate. It is built to break the continuity of a monitoring trail that expects one channel at a time.

Stage 4: Exit (Hours 2-24)

Money leaves through remittance corridors, digital asset platforms, or trusted networks, often before any batch report or investigation has a chance to catch up.

Indonesia’s real-time payment rails make the exit phase especially fast. With BI-FAST (Bank Indonesia’s real-time interbank transfer system) and QRIS enabling instant, 24/7 transfers, the window between activation and exit has compressed from days to minutes. Bank Indonesia

Why your current monitoring misses this

Most Indonesian banks have strong transaction monitoring systems. But they have three structural gaps:

Gap 1: Channel isolation

E-money wallet movement and core banking movement get reported separately. A suspicious pattern only becomes visible when the events are connected.

Gap 2: Static risk profiles

A customer’s risk rating is set at onboarding and rarely updated based on behavioural changes. A low-risk customer can become high-risk within hours, but the system still treats them as low-risk.

Gap 3: Batch-based detection

Many banks still rely on end-of-day or end-of-week batch reports. A mule account that would previously have been caught in batch controls can now move funds across banks, e-money wallets and remittance channels before the batch ever fires.

What OJK and Bank Indonesia are watching

OJK and Bank Indonesia are not yet framing this specifically as a dormancy problem. But enforcement focus is shifting.

Based on current supervisory discussion and international pressure on Indonesia’s AML effectiveness, banks should prepare for:

  • Continuous risk assessment extending well past initial CDD, requiring banks to reassess account usage against declared purpose throughout the relationship
  • Dormancy-to-activation monitoring becoming an explicit supervisory expectation, flagging accounts that jump from dormant to high-activity after 60 or more days of inactivity
  • Cross-channel visibility rules requiring banks to see core banking, e-money and card activity simultaneously rather than reconciling it in batch
  • Predefined red flags for known mule typologies, such as dormancy followed by rapid outflow or simultaneous multi-channel movement

Banks that build these controls now, ahead of formal guidance, put themselves in a stronger position with regulators.

Five actions your fraud desk should take now

  1. Build dormancy profiles by customer segment

A student account sitting untouched for six months may be normal. A business account doing the same may not be. Calibrate dormancy thresholds by segment, product type and historical behaviour.

  1. Create an activation risk signal

When a dormant account suddenly activates, assess:

  • How long was the account inactive?
  • What triggered the first activity?
  • Who sent the money?
  • How quickly did the funds leave?
  • Were several channels involved?
  • Does the activity match the customer’s declared purpose?
  1. Connect channels in real time

Set up real-time correlation: when an account receives funds in core banking and shows outflow activity in a mobile wallet within two hours, treat that as one suspicious pattern, not two separate transactions.

  1. Validate incoming fund sources

Mule accounts often receive funds from recently compromised business accounts or known investment-scam recipients. Maintain a live list of flagged source accounts and review any dormant account that receives funds from them upon activation.

  1. Reconcile activity with purpose

Compare how an account is actually used against what the customer declared at onboarding. A customer who said they were opening an account for e-commerce purchases should not suddenly be receiving large transfers from multiple unrelated sources.

The business case for early action

Banks that move first on post-onboarding risk reassessment and cross-channel correlation stand to gain:

Regulatory credibility

When OJK eventually issues guidance on these controls, early movers will have months of operational data and maturity to show for it.

Reputation and trust

Banks that act early get to define their own reputation on this issue with customers and regulators alike.

Operational efficiency

A fraud desk that can identify mule accounts at activation, rather than weeks into the layering cycle, files more actionable Suspicious Transaction Reports (STRs) with Indonesia’s Financial Transaction Reports and Analysis Centre (PPATK).

How Clari5 approaches this problem

Clari5, a Perfios company, has been recognised across key industry benchmarks. In Chartis Research‘s Enterprise & Payment Fraud Solutions 2026 Quadrant Report, Clari5 was named a Category Leader across all three quadrants. In 2025, it was also named a Category Leader across four Chartis fraud quadrants, including Identity & Verification (ID&V). Most recently, Clari5 secured a Top 50 position in the Chartis Research FCC50 2026 rankings.

Our approach is built around Indonesia’s dormancy-and-activation reality:

  • Persistent customer risk profiles that update through every lifecycle stage, from onboarding through dormancy, activation and active use
  • Real-time cross-channel correlation across core banking, card and e-money channels
  • Behavioural machine learning trained on known mule-activation typologies
  • Post-activation risk reassessment that compares current usage against declared purpose
  • Explainable, auditable models so compliance and internal audit teams can trace why an account was re-rated as high risk

This is not about replacing your existing controls. It is about extending the same rigor you apply at onboarding further into the account lifecycle.

Key takeaway for bank leaders

Your onboarding controls have already done their job well. The next place to focus is the dormancy-to-activation transition, where purpose drift actually shows up.

A clean account can become a risky account.

The bank that detects that change early can protect the customer, preserve more funds and give investigators a better chance of disrupting the wider network.

That is the real purpose of lifecycle-based fraud and AML monitoring: not to distrust customers, but to protect legitimate financial activity when criminal behaviour enters the system.

Frequently asked questions

What is a mule account in Indonesia?

A mule account is a bank or payment account used to receive, hold or transfer funds obtained through fraud, scams or other financial crime. The account may belong to a recruited participant, a victim whose credentials were compromised or a person knowingly allowing others to use the account.

Why do mule accounts pass KYC checks?

KYC verifies identity and customer information at onboarding. It does not always reveal how an account will be used months later. Continuous monitoring is needed to identify changes in behaviour, purpose and transaction patterns.

What is dormancy-to-activation monitoring?

It is the monitoring of accounts that remain inactive for a defined period and then begin transacting suddenly or intensively. The risk assessment considers the dormancy period, first transaction, source of funds, velocity, beneficiaries and cross-channel activity.

How should Indonesian banks detect mule accounts?

Banks should combine customer lifecycle data, transaction monitoring, device and channel information, network relationships, external intelligence from IASC and investigator review. No single threshold is sufficient for every customer segment.

How does IASC support scam response in Indonesia?

IASC provides a coordinated channel for handling reports of financial scams and supporting the blocking and recovery of suspected funds. Customers should report scams through the official IASC website as soon as possible. CNBC Indonesia

Does stronger mule-account detection require blocking more customers?

Not necessarily. Risk-based controls can use step-up verification, temporary holds, customer confirmation and targeted investigation instead of automatically blocking every unusual transaction.

What will OJK expect from banks on mule accounts?

While OJK has not issued specific dormancy guidance, banks should prepare for closer supervisory attention to continuous customer risk assessment, behavioural monitoring and speed of response. Any future OJK or Bank Indonesia guidance should be treated as authoritative when issued.

How can banks measure success in mule-account detection?

Track time from activation to detection, time from detection to intervention, funds prevented from leaving, recovery rate, repeat use of linked accounts, number of related accounts identified, quality and timeliness of STR escalation, and customer impact and false-positive rates.