Glossary of Financial Crime Compliance / AI Terms

Regulators & Authorities

• Office of the Comptroller of the Currency (OCC) – Regulates all US banks.

• Monetary Authority of Singapore (MAS) – Regulates financial institutions in Singapore.

• FinCEN (Financial Crimes Enforcement Network) – US Treasury bureau responsible for combatting money laundering and terrorist financing.

• Financial Action Task Force (FATF) – Global money laundering and terrorist financing watchdog. Issues blacklists/grey lists of high-risk jurisdictions.

• United Nations Office on Drugs and Crime (UNODC) – UN body leading global action against illicit drugs, crime, and terrorism.

• Financial Conduct Authority (FCA) – UK regulator of financial services.

• European Banking Authority (EBA) – Oversees banking supervision across the EU.

• Basel Committee on Banking Supervision (BCBS) – International forum for setting banking supervision standards.

• OFAC (Office of Foreign Assets Control) – US authority administering sanctions programmes.

Financial Crime Compliance Terms

• AML (Anti-Money Laundering)  – Laws, regulations, and procedures aimed at preventing criminals from disguising illicit funds as legitimate income.

• KYC (Know Your Customer) – Due diligence processes to verify customer identity, suitability, and risks.

• CDD (Customer Due Diligence) – Standard checks carried out on clients during onboarding.

• PEPs (Politically Exposed Persons) – Individuals with prominent public functions who present higher corruption risks.

• Sanctions Screening – Checking names, entities, and transactions against sanctions lists.

• Suspicious Activity Report (SAR) – A formal report filed by financial institutions to regulators when they detect suspicious activity that may involve money laundering, fraud, terrorist financing, or other financial crimes.

• Transaction Monitoring – Ongoing surveillance of financial transactions for unusual patterns.

• Correspondent Banking – Relationships that allow cross-border payments and are often high risk for money laundering.

• Suspicious Transaction Report (STR) - A mandatory report filed when a transaction itself is considered suspicious or inconsistent with a customer’s known profile or business activities.

• Transaction Screening – Real-time filtering of payments or trades against sanctions, watchlists, or risk indicators before execution.

• Name Screening – Checking customer or counterparty names against sanctions, PEP, and adverse media lists for potential risk matches.

• Entity Resolution – The process of identifying, linking, and consolidating records that refer to the same individual or organisation across data sources.

• FRAML – The convergence of Fraud and Anti-Money Laundering controls into a unified approach to detect and prevent financial crime.

• ISO 20022 – A global financial messaging standard that enables richer, structured data exchange in payments and securities.

• Trade Surveillance – Monitoring trading activity to detect market abuse, insider dealing, or manipulative behaviors.

• Straight Through Processing (STP) – In FCC, STP refers to the automatic clearance of alerts or transactions without human intervention, when they meet predefined rules or deterministic logic. This allows low-risk cases to flow through seamlessly, reducing manual workload while ensuring compliance standards are upheld.

• Trade Based Money Laundering (TBML) - The process of disguising the proceeds of crime and moving value through trade transactions to legitimise illicit funds. Criminals exploit the complexity and volume of international trade to obscure the origins of money.

AI & Technology in Compliance

• Automated Alert Adjudication – The use of AI or rule-based systems to automatically review and resolve alerts generated by monitoring or screening tools, replicating human decision logic to clear false positives or escalate true matches, while maintaining auditability and compliance standards.

• Natural Language Processing (NLP) – AI technique for understanding and analysing human language in alerts, reports, or regulatory texts.

• Natural Language Output (NLO) – When the system’s reasoning or results are presented in plain text.

• Machine Learning (ML) – AI methods enabling systems to learn from data and improve detection of suspicious activity.

• Explainable AI (XAI) – AI models designed to be transparent, showing how decisions (e.g., alerts) were reached.

• False Positives – Legitimate transactions incorrectly flagged as suspicious.

• False Negative – A missed detection in financial crime compliance, where a truly suspicious activity, transaction, or name match is not flagged by the monitoring or screening system.

• Entity Resolution – Process of matching and consolidating records referring to the same person or company.

• Graph Analytics – Network analysis techniques used to detect hidden links between entities in financial crime networks.

• Generative AI – AI models that can generate text, summaries, or insights to assist compliance analysts.

• Human-in-the-Loop (HITL) – A model governance approach where human experts remain actively involved in reviewing, validating, or overriding automated system decisions, ensuring accuracy, accountability, and regulatory compliance.

• Model Risk Management (MRM) – Framework for controlling risks associated with deploying AI/ML models in regulated environments.

• Supervised vs. Unsupervised Learning – ML methods: supervised uses labelled data (e.g., past SARs), unsupervised identifies anomalies without labels.

• Deterministic Decisioning / Model – A rules-based system that always produces the same output given the same input.

• Model Validation – Independent testing to confirm a model’s accuracy, reliability, and regulatory compliance.

• Model Drift – Degradation of model performance over time due to changes in data, behavior, or environment.

• Model Explainability (XAI) – Techniques that make AI/ML model decisions transparent and understandable to humans.

• Robotic Process Automation (RPA) – Software that automates repetitive, rule-based tasks without human intervention.

• Hallucination – When an AI system generates outputs that are incorrect, fabricated, or not grounded in the data.