Compliance Hub

Ensure COAF Compliance | Financial Activities Control in Brazil

Site Logo
Tookitaki
4 min
read

The Council for Financial Activities Control or Conselho de Controle de Atividades Financeiras  (COAF) is Brazil’s national financial intelligence unit, responsible for combating money laundering in the country.

The COAF receives and analyses information on financial transactions that have potential links to financial crimes including money laundering. Following its analysis, the council disseminates the results to the competent authorities when there are grounds to suspect money laundering or terrorist financing.

 

The Origin and Development of COAF

The COAF was created as part of Brazil’s Anti-Money Laundering Law passed in March 1998. Working under the Ministry of Finance, the council had tasks including regulating, applying administrative sanctions, receiving pertinent information, and examining and identifying suspicious activities related to money laundering.

The COAF’s structure, laws and operations were defined in subsequent legislations enacted between October and December 1998.

In August 2019, the council was renamed to Financial Intelligence Unit (Unidade de Inteligência Financeira in Portuguese) by President Jair Bolsonaro and was brought under the control of the Central Bank of Brazil. However, the National Congress of Brazil overruled the decision in January 2020 and decided to roll back the name to the Council for Financial Activities Control (COAF).

 

What Are the Responsibilities of COAF?

The following are the responsibilities of COAF:

  • Ensuring that Brazilian financial institutions operate in compliance with the country’s anti-money laundering (AML) regulations by implementing suitable policies and controls
  • Assessing the AML compliance programmes of firms operating in Brazil and analysing the suspicious activity reports (SARs) that they submit
  • Representing Brazil on international AML platforms such as the Financial Action Task Force (FATF)
  • Conducting investigations and disseminating key information to relevant law enforcement authorities when the agency detects violations
  • Providing the public with information and statistics relating to financial crime and compliance with various publications

 

How to Ensure COAF Compliance?

To comply with the COAF, firms in Brazil should have proper financial crime compliance programmes in line with the country’s regulations. Processes such as Customer Due Diligence (CDD), Watchlist Screening, Transaction Screening and Transaction Monitoring ensure that companies comply with regulations and avoid significant fines.

AML compliance programmes should have the following key components:

 

  • Customer Due Diligence: Through this process, firms verify the identities of their customers while onboarding and throughout the customer lifecycle to build accurate risk profiles.
  • Transaction Monitoring: In this process, firms monitor customer transactions for unusual activities (eg. abnormal transaction patterns, transactions that differ from risk profiles, or transactions with high-risk countries).
  • Watchlist Screening: This process ensures the screening of customers against various local and international watchlists (eg. sanctions, politically exposed persons) so that financial institutions can take necessary controls while doing business with high-risk customers.
  • Transaction Screening: Here, financial institutions screen third-party payment details against various watchlists so that they can avoid the risk of channelling funds into the wrong hands.

 

How Can Tookitaki Help Financial Institutions in Brazil?

Headquartered in Singapore, Tookitaki is a regulatory technology company offering financial crime detection and prevention to some of the world's leading banks and fintech companies to help them transform their anti-financial crime and compliance technology needs. Founded in November 2014, the Company employs over 100 people across Asia, Europe, and the US.

Fighting financial crime needs to be a collective effort through centralised intelligence-gathering. The Anti-Financial Crime (AFC) Ecosystem includes a network of experts and provides a platform for the experts to create a knowledge base to share financial crime scenarios.

This collective intelligence is the ability of a large group of AFC experts to pool their knowledge, data, and skills in order to tackle complex problems related to financial crime and pursue innovative ideas.

The AFC ecosystem is a game changer since it helps remove the information vacuum created by siloed operations. Our network of experts includes risk advisers, legal firms, AFC specialists, consultancies, and financial institutions from across the globe.

Tookitaki’s Anti-Money Laundering Suite (AMLS) covers the entire customer onboarding and ongoing processes through its Transaction Monitoring, Smart Screening, Customer Risk Scoring, and Case Manager. Together they provide holistic risk coverage, sharper detection, and significant effort reduction in managing false alerts.

Tookitaki's solutions work in tandem and help our stakeholders widen their view of risk from an internal one to an industry-wide one across organizations and borders. Moreover, they can do so without compromising privacy and security.

Apart from necessary human resources, banks and financial services should have technological resources to carry out their AML compliance activities and duties effectively.  Our offerings manage the end-to-end of AML compliance programmes and help improve the efficiency of the AML compliance team and better mitigate compliance risk.

Speak to one of our experts today to understand how our solutions help your compliance teams to ensure future-proof compliance programmes.

 

By submitting the form, you agree that your personal data will be processed to provide the requested content (and for the purposes you agreed to above) in accordance with the Privacy Notice

success icon

We’ve received your details and our team will be in touch shortly.

In the meantime, explore how Tookitaki is transforming financial crime prevention.
Learn More About Us
Oops! Something went wrong while submitting the form.

Ready to Streamline Your Anti-Financial Crime Compliance?

Our Thought Leadership Guides

Blogs
14 Aug 2025
5 min
read

Smarter Investigations: The Rise of AML Investigation Tools in Australia

In the battle against financial crime, the right AML investigation tools turn data overload into actionable intelligence.

Australian compliance teams face a constant challenge — growing transaction volumes, increasingly sophisticated money laundering techniques, and tighter AUSTRAC scrutiny. In this environment, AML investigation tools aren’t just nice-to-have — they’re essential for turning endless alerts into fast, confident decisions.

Talk to an Expert

Why AML Investigations Are Getting Harder in Australia

1. Explosion of Transaction Data

With the New Payments Platform (NPP) and cross-border corridors, institutions must monitor millions of transactions daily.

2. More Complex Typologies

From mule networks to shell companies, layering techniques are harder to detect with static rules alone.

3. Regulatory Expectations

AUSTRAC demands timely and accurate Suspicious Matter Reports (SMRs). Delays or incomplete investigations can lead to penalties and reputational damage.

4. Resource Constraints

Skilled AML investigators are in short supply. Teams must do more with fewer people — making efficiency critical.

What Are AML Investigation Tools?

AML investigation tools are specialised software platforms that help compliance teams analyse suspicious activity, prioritise cases, and document findings for regulators.

They typically include features such as:

  • Alert triage and prioritisation
  • Transaction visualisation
  • Entity and relationship mapping
  • Case management workflows
  • Automated reporting capabilities

Key Features of Effective AML Investigation Tools

1. Integrated Case Management

Centralise all alerts, documents, and investigator notes in one platform.

2. Entity Resolution & Network Analysis

Link accounts, devices, and counterparties to uncover hidden connections in laundering networks.

3. Transaction Visualisation

Graph-based displays make it easier to trace fund flows and identify suspicious patterns.

4. AI-Powered Insights

Machine learning models suggest likely outcomes, surface overlooked anomalies, and flag high-risk entities faster.

5. Workflow Automation

Automate repetitive steps like KYC refresh requests, sanctions re-checks, and document retrieval.

6. Regulator-Ready Reporting

Generate Suspicious Matter Reports (SMRs) and audit logs that meet AUSTRAC’s requirements.

ChatGPT Image Aug 13, 2025, 12_27_28 PM

Why These Tools Matter in Australia’s Compliance Landscape

  • Speed: Fraud and laundering through NPP happen in seconds — investigations need to move just as fast.
  • Accuracy: AI-driven tools reduce false positives, ensuring analysts focus on real threats.
  • Compliance Assurance: Detailed audit trails prove that due diligence was carried out thoroughly.

Use Cases in Australia

Case 1: Cross-Border Layering Detection

An Australian bank flagged multiple small transfers to different ASEAN countries. The AML investigation tool mapped the network, revealing links to a known mule syndicate.

Case 2: Crypto Exchange Investigations

AML tools traced a high-value Bitcoin-to-fiat conversion back to an account flagged in a sanctions database, enabling rapid SMR submission.

Advanced Capabilities to Look For

Federated Intelligence

Access anonymised typologies and red flags from a network of institutions to spot emerging threats faster.

Embedded AI Copilot

Assist investigators in summarising cases, recommending next steps, and even drafting SMRs.

Scenario Simulation

Test detection scenarios against historical data before deploying them live.

Spotlight: Tookitaki’s FinCense and FinMate

FinCense integrates investigation workflows directly into its AML platform, while FinMate, Tookitaki’s AI investigation copilot, supercharges analyst productivity.

  • Automated Summaries: Generates natural language case narratives for internal and regulatory reporting.
  • Risk Prioritisation: Highlights the highest-risk cases first.
  • Real-Time Intelligence: Pulls in global typology updates from the AFC Ecosystem.
  • Full Transparency: Glass-box AI explains every decision, satisfying AUSTRAC’s audit requirements.

With FinCense and FinMate, Australian institutions can cut investigation times by up to 50% — without compromising quality.

Conclusion: From Data to Decisions — Faster

The volume and complexity of alerts in modern AML programmes make manual investigation unsustainable. The right AML investigation tools transform scattered data into actionable insights, helping compliance teams stay ahead of both criminals and regulators.

Pro tip: Choose tools that not only investigate faster, but also learn from every case — making your compliance programme smarter over time.

Smarter Investigations: The Rise of AML Investigation Tools in Australia
Blogs
13 Aug 2025
5 min
read

Smarter Defences: How Machine Learning is Transforming Fraud Detection in Philippine Banking

Fraud in banking has never been faster, smarter, or more relentless — and neither have the defences.

In the Philippines, the rapid rise of digital banking, mobile wallets, and instant payments has created unprecedented opportunities for growth — and for fraudsters. From account takeovers to synthetic identity scams, financial institutions are under constant attack. Traditional rule-based detection systems, while useful, are no longer enough. Enter machine learning (ML) — the technology redefining fraud detection by spotting suspicious activity in real time and adapting to new threats before they cause damage.

Talk to an Expert

The Growing Fraud Threat in Philippine Banking

Digital banking adoption in the Philippines has surged in recent years, driven by initiatives like the BSP’s Digital Payments Transformation Roadmap and the expansion of fintech services. While these advancements boost financial inclusion, they also open the door to fraud.

According to the Bankers Association of the Philippines, reported cyber fraud incidents have increased steadily, with phishing, account takeover (ATO), and card-not-present (CNP) fraud among the top threats.

Key trends include:

  • Instant payment exploitation: Fraudsters leveraging PESONet and InstaPay for rapid fund transfers.
  • Social engineering scams: Convincing victims to disclose personal and banking details.
  • Cross-border fraud networks: Syndicates funnelling illicit funds via multiple jurisdictions.

In this environment, speed, accuracy, and adaptability are critical — qualities where ML excels.

Why Traditional Fraud Detection Falls Short

Rule-based fraud detection systems rely on predefined scenarios (e.g., flagging transactions over a certain threshold or unusual logins from different IP addresses). While they can catch known patterns, they struggle with:

  • Evolving tactics: Fraudsters quickly adapt once they know the rules.
  • False positives: Too many alerts waste investigator time and frustrate customers.
  • Lack of contextual awareness: Rules can’t account for the nuances of customer behaviour.

This is where machine learning transforms the game.

How Machine Learning Enhances Fraud Detection

1. Pattern Recognition Beyond Human Limits

ML models can process millions of transactions in real time, identifying subtle anomalies in behaviour — such as unusual transaction timing, frequency, or geolocation.

2. Continuous Learning

Unlike static rules, ML systems learn from new data. When fraudsters switch tactics, the model adapts, ensuring defences stay ahead.

3. Reduced False Positives

ML distinguishes between legitimate unusual behaviour and true fraud, cutting down on unnecessary alerts. This not only saves resources but improves customer trust.

4. Predictive Capability

Advanced algorithms can predict the likelihood of a transaction being fraudulent based on historical and behavioural data, enabling proactive intervention.

ChatGPT Image Aug 13, 2025, 12_05_50 PM

Key Machine Learning Techniques in Banking Fraud Detection

Supervised Learning

Models are trained using labelled datasets — past transactions marked as “fraud” or “legitimate.” Over time, they learn the characteristics of fraudulent activity.

Unsupervised Learning

Used when there’s no labelled data, these models detect outliers and anomalies without prior examples, ideal for spotting new fraud types.

Reinforcement Learning

The system learns by trial and error, optimising decision-making as it receives feedback from past outcomes.

Natural Language Processing (NLP)

NLP analyses unstructured data such as emails, chat messages, or KYC documents to detect potential fraud triggers.

Real-World Fraud Scenarios in the Philippines Where ML Makes a Difference

  1. Account Takeover (ATO) Fraud – ML flags login attempts from unusual devices or geolocations while analysing subtle session behaviour patterns.
  2. Loan Application Fraud – Models detect inconsistencies in credit applications, cross-referencing applicant data with external sources.
  3. Payment Mule Detection – Identifying suspicious fund flows in real time, such as rapid inbound and outbound transactions in newly opened accounts.
  4. Phishing-Driven Transfers – Correlating unusual fund movement with compromised accounts reported across multiple banks.

Challenges in Implementing ML for Fraud Detection in the Philippines

  • Data Quality and Availability – ML models need vast amounts of clean, structured data. Gaps or inaccuracies can reduce effectiveness.
  • Regulatory Compliance – BSP regulations require explainability in AI models; “black box” ML can be problematic without interpretability tools.
  • Talent Gap – Limited availability of data science and ML experts in the local market.
  • Integration with Legacy Systems – Many Philippine banks still run on legacy infrastructure, complicating ML deployment.

Best Practices for Deploying ML-Based Fraud Detection

1. Start with a Hybrid Approach

Combine rule-based and ML models initially to ensure smooth transition and maintain compliance.

2. Ensure Explainability

Use explainable AI (XAI) frameworks so investigators and regulators understand why a transaction was flagged.

3. Leverage Federated Learning

Share intelligence across institutions without exposing raw data, enhancing detection of cross-bank fraud schemes.

4. Regular Model Retraining

Update models with the latest fraud patterns to stay ahead of evolving threats.

5. Engage Compliance Early

Work closely with risk and compliance teams to align ML use with BSP guidelines.

The Tookitaki Advantage: The Trust Layer to Fight Financial Crime

Tookitaki’s FinCense platform is built to help Philippine banks combat fraud and money laundering with Agentic AI — an advanced, explainable AI framework aligned with global and local regulations.

Key benefits for fraud detection in banking:

  • Real-time risk scoring on every transaction.
  • Federated intelligence from the AFC Ecosystem to detect emerging fraud typologies seen across the region.
  • Lower false positives through adaptive models trained on both local and global data.
  • Explainable decision-making that meets BSP requirements for transparency.

By combining advanced ML techniques with collaborative intelligence, FinCense gives banks in the Philippines the tools they need to protect customers, meet compliance standards, and reduce operational costs.

Conclusion: Staying Ahead of the Curve

Fraudsters in the Philippines are becoming more sophisticated, faster, and harder to trace. Relying on static, rules-only systems is no longer an option. Machine learning empowers banks to detect fraud in real time, reduce false positives, and adapt to ever-changing threats — all while maintaining compliance.

For institutions aiming to build trust in a rapidly digitising market, the path forward is clear: invest in ML-powered fraud detection now, and make it a core pillar of your risk management strategy.

Smarter Defences: How Machine Learning is Transforming Fraud Detection in Philippine Banking
Blogs
13 Aug 2025
5 min
read

Stopping Fraud in Its Tracks: The Future of Transaction Fraud Detection in Singapore

Fraud doesn’t knock—it slips through unnoticed until it’s too late.

As digital payments accelerate across Singapore, financial institutions face a mounting challenge: detecting fraudulent transactions in real time, without slowing down legitimate users. From phishing scams and mule accounts to synthetic identities and account takeovers, transaction fraud has become smarter, faster, and harder to catch.

This blog explores how transaction fraud detection is evolving in Singapore, the gaps still present in legacy systems, and how AI-driven tools are helping financial institutions fight back.

Talk to an Expert

Why Transaction Fraud Detection Is Critical in Singapore

Singapore’s position as a fintech hub comes with exposure to increasingly sophisticated fraud schemes. According to the Singapore Police Force, scam-related crimes in 2024 accounted for over 70% of all crimes reported, with transaction fraud and unauthorised transfers making up a large portion of the losses.

The government’s drive for real-time payments — from PayNow to FAST — adds pressure on banks and fintechs to detect fraud instantly, without delaying genuine transactions.

Missed fraud isn’t just a financial risk — it erodes trust. And in Singapore’s tightly regulated environment, trust is everything.

Types of Transaction Fraud Facing Financial Institutions

Understanding the tactics fraudsters use is the first step toward stopping them. In Singapore, common forms of transaction fraud include:

1. Account Takeover (ATO)

Fraudsters use stolen credentials to gain control over an account and initiate transfers, bill payments, or cash withdrawals — often within minutes.

2. Social Engineering Scams

Victims are tricked into authorising payments themselves under false pretences — for example, investment scams, job scams, or fake relationships.

3. Money Muling

Fraudsters use mule accounts — often belonging to unsuspecting individuals — to route stolen or laundered funds through multiple hops.

4. Real-Time Payment Exploits

With instant transfer systems, once funds are sent, they’re often impossible to recover. Fraudsters exploit this urgency and invisibility.

5. Business Email Compromise (BEC)

Corporate payments are manipulated through phishing or spoofing attacks, redirecting funds to illicit accounts under false vendor names.

ChatGPT Image Aug 13, 2025, 11_14_07 AM

Challenges in Transaction Fraud Detection

Despite investment in fraud controls, many Singaporean financial institutions still face persistent roadblocks:

1. High False Positives

Basic rules-based systems raise alerts for normal user behaviour, overwhelming fraud teams and increasing friction for genuine customers.

2. Lack of Real-Time Detection

Many systems rely on batch processing or delayed scoring, leaving gaps for fraudsters to exploit instant payment rails.

3. Inability to Detect Novel Patterns

Fraudsters constantly change tactics. Systems that only recognise known fraud signatures are easily bypassed.

4. Poor Cross-Border Visibility

Singapore is deeply integrated into global financial flows. A lack of insight into transaction trails beyond borders makes it harder to detect layered laundering and syndicated fraud.

What Effective Transaction Fraud Detection Looks Like Today

Modern fraud detection is about being predictive, not just reactive. Here's what best-in-class solutions offer:

AI + Machine Learning

Rather than using only static rules, intelligent systems learn from historical patterns, adapt to new behaviours, and improve accuracy over time.

Behavioural Profiling

These systems build user profiles based on login patterns, spending habits, device data, and more — flagging anything outside the norm in real time.

Network Analysis

Sophisticated fraud often involves mule networks or linked entities. Graph analysis helps identify suspicious linkages between accounts.

Federated Intelligence Sharing

Platforms like Tookitaki’s AFC Ecosystem allow institutions to benefit from typologies and red flags contributed by others — without sharing sensitive data.

Explainable AI

Regulators require transparency. Solutions must explain why a transaction was flagged, not just that it was.

How Tookitaki Is Powering Smarter Fraud Detection

Tookitaki’s FinCense platform is purpose-built to detect transaction fraud in real time. Here’s how it helps Singapore-based institutions stay ahead:

  • Agentic AI Framework: Modular AI agents continuously scan transactions, user behaviour, and risk context to identify fraud patterns — even emerging ones.
  • Scenario-Based Detection: Leverages real-world fraud scenarios from the AFC Ecosystem, including scams unique to Southeast Asia like fake job recruitment and QR-enabled mule layering.
  • Real-Time Simulation & Threshold Optimisation: Before deploying rules, institutions can simulate detection impact to reduce false positives.
  • Smart Disposition Engine: AI-generated summaries assist investigators by surfacing key risk insights for flagged transactions.
  • Federated Learning: Combines privacy-preserving AI with community-sourced intelligence for faster, more adaptive detection.

Whether you’re a digital bank, a payment gateway, or a traditional financial institution, FinCense provides the flexibility, speed, and accuracy needed for the Singaporean fraud landscape.

Key Strategies for Singaporean Firms to Strengthen Fraud Defences

1. Upgrade From Rule-Based to Hybrid Systems

A combination of dynamic rules and machine learning provides greater precision and adaptability.

2. Focus on Early Detection

Identify mule accounts, layered transfers, and behaviour anomalies before the fraud is completed.

3. Enable Seamless Analyst Workflows

Reduce alert fatigue with AI-driven prioritisation and investigation summaries.

4. Join Intelligence-Sharing Networks

Collaborate with platforms like the AFC Ecosystem to keep up with evolving fraud typologies.

5. Design for Real-Time Action

Ensure that fraud decisions can be made in milliseconds — and tie detection systems directly to block/hold actions.

Conclusion: Fraudsters Are Getting Smarter. Are You?

In Singapore’s fast-moving financial ecosystem, transaction fraud detection is no longer just a compliance function — it’s a competitive advantage.

Banks and fintechs that invest in modern, intelligent fraud prevention are not only protecting their bottom line — they’re protecting their brand and customer relationships.

📌 The future of fraud detection is proactive, predictive, and powered by community-led intelligence. Don’t just keep up — get ahead.

Stopping Fraud in Its Tracks: The Future of Transaction Fraud Detection in Singapore