Financial institutions operating in Argentina must manage the risks of money laundering, terrorist financing, and the proliferation of weapons of mass destruction in accordance with their activities, risk profiles, and applicable regulations. For financial institutions and foreign exchange entities, UIF Resolution 14/2023 establishes a risk-based approach and was amended by UIF Resolution 199/2024 [1]. Virtual asset service providers are subject to a specific regulatory framework under UIF Resolution 49/2024 [2].
The joint FATF/GAFILAT Mutual Evaluation was adopted in October 2024 and published in December of that year [3]. The report acknowledged improvements in Argentina's regulatory framework while identifying areas where the country needs to strengthen the effectiveness of its system. For institutions, this environment makes it particularly important to maintain documented controls that are proportionate to their risks and subject to regular review.
For Compliance, Risk, and Technology teams, the challenge is to maintain transaction monitoring, analyze deviations from customer profiles, and investigate alerts with sufficient context. The question is not whether a particular tool replaces the existing compliance framework, but rather how it can improve detection and investigative processes within each institution's governance model.
From rule-based monitoring to adaptive risk management
Rule-based scenarios remain useful for implementing defined controls and detecting known patterns. At the same time, institutions can evaluate advanced analytics and machine learning to identify behavioral changes, relationships between counterparties, and anomalies that warrant further review.
A monitoring solution can combine model-generated signals, configurable rules, and institution-specific risk criteria. The operational objective is to prioritize alerts with sufficient context so that analysts can understand which behaviors triggered the alert, what data supports it, and which review steps are required.
What artificial intelligence can deliver
- Behavioral analysis: by comparing observed activity with historical context and available customer profiles, models can identify deviations that require human review.
- Relationships and networks: analyzing connections between customers, accounts, and counterparties can provide additional context about patterns distributed across multiple entities, where such data is available and can be used in accordance with applicable regulations.
- Prioritization and explainability: an effective alert should include risk factors, related transactions, and a clear, understandable explanation that supports review, decision documentation, and auditing. Models should also avoid unexplainable "black-box" architectures that cannot be adequately interpreted by auditors and regulators.
- Investigation support: AI tools can help organize evidence, summarize background information, and prepare draft analyses. Their scope of use must be defined by the contracted product and the institution's approved procedures.
Cloud infrastructure, integration, and institutional responsibility
The SaaS or cloud delivery model optimizes Total Cost of Ownership (TCO) and enhances operational agility by redirecting capital resources (CapEx) toward improving model accuracy. Through lightweight, secure APIs, these solutions enable transactional platforms to connect to an intelligence engine without modifying legacy core systems.
For institutions supervised by the Central Bank of Argentina (BCRA), outsourcing arrangements, including cloud service providers, require risk management practices aligned with the criticality of the service under current supervisory guidelines [4]. Among other considerations, institutions must address business continuity and recovery, provider selection and oversight, contractual responsibilities, and control over outsourced activities. Technology can support the compliance program, but responsibility for its governance remains with the institution.
The specific architecture, whether SaaS, institution-managed cloud, or on-premises infrastructure, must be described according to the solution offered and the design agreed upon with each institution. At a minimum, the assessment should cover integration, data access, security, business continuity, auditing, subcontracting, and exit strategies.
Investigations and suspicious transaction reporting
AI can assist in preparing investigations by gathering data, organizing evidence, highlighting inconsistencies, and proposing draft narratives. Decisions regarding whether a transaction should be reported, as well as approval of the report, must remain within the procedures and responsibilities of the reporting entity.
In Argentina, the Financial Intelligence Unit (UIF) requires Suspicious Transaction Reports (ROS) to be properly substantiated and to include the reasons why the reporting entity considers the transaction suspicious [5].
Before claiming that a platform automates ROS submissions, it is necessary to verify that the workflow, format, and any integrations comply with current UIF requirements.
Strategic value for banks and fintechs
Modernizing compliance through SaaS serves as a direct enabler for different segments of the financial sector:
- Traditional banks: it enables institutions to modernize their analytical capabilities without replacing their underlying core banking systems, while generating operational efficiencies by reducing noise in initial alert screening.
- Neobanks, digital wallets, and digital challengers: it provides horizontal scalability to accommodate rapid user growth and enables institutions to expand their customer portfolios while complying with UIF and FATF regulations, without proportionally increasing fixed staffing costs.
Globally proven, now operating in Argentina
This architecture is more than a technical promise. It has been tested and validated by ThetaRay, a pioneering company in the development of machine learning algorithms specialized in financial risk detection.
ThetaRay's AI platform is part of the global operations of some of the world's largest financial groups, including Spanish and French banking institutions with hundreds of billions of dollars in assets under management (AUM).
The multinational deployment of these SaaS models has demonstrated that high detection accuracy can be combined with strict regulatory compliance.
Today, this same technology is operating within leading banking institutions in Argentina, with another major local financial institution expected to join soon.
The value of local implementation
Implementation requires translating each institution's compliance priorities into scenarios, data, rules, risk profiles, and investigation workflows that reflect its operations.
This includes validating product and channel coverage, data availability, and decision-documentation processes for Compliance, Risk, Technology, and Audit teams.
INDIGO IA supports institutions in evaluating, integrating, and deploying artificial intelligence solutions, working in coordination with the technology provider and the client's internal teams.
Configuration, results, and impact indicators must be designed and calibrated to the specific characteristics and transaction volumes of each institution, with validation based on data and metrics agreed upon for each project.
AI adoption in compliance is best evaluated through measurable objectives: alert quality and usefulness, investigation times, traceability, risk coverage, and integration stability.
These measurements help institutions determine where technology adds value and how to maintain human oversight and regulatory control.
Regulatory and contextual references
- UIF Resolution 14/2023, updated text, including amendments introduced by UIF Resolution 199/2024.
- UIF Resolution 49/2024, regulatory framework applicable to virtual asset service providers.
- FATF/GAFILAT, Argentina Mutual Evaluation 2024, adopted in October and published in December 2024.
- Central Bank of Argentina (BCRA), Supervisory Guidelines, including criteria related to third-party risk management and cloud services.
- UIF, Frequently Asked Questions on Transaction Reporting and Suspicious Transaction Reports (ROS), including reporting justifications and applicable deadlines.
