ACAMS Today, March-May 2025 | Page 51

At the same time , the challenges linked to the widescale adoption of AI are equally as often voiced . Some are linked to skepticism of stakeholders due to the lack of transparency attached to AI models as well as reservations around data protection and privacy .
As early as 2017 , McKinsey & Company published a study 1 outlining the various inefficiencies it had identified within compliance teams , which included fragmented efforts , manual processes and mountains of data . 2 The adoption of AI has been slow , however , and only very recently have large banks started to explore and test the integration of AI into their AFC programs . Fintechs have been bolder in that regard . 3 Fast and tailored adoption can be expected in the coming years across the financial sector given that the European Union ( EU ) has introduced data protection and AI regulation in 2018 and 2024 , respectively , providing the necessary assurance through a robust regulatory framework .
This article will assess how AI and data privacy regulation have sought to address the concerns linked to the adoption of AI , and how obliged entities that consider the appropriate implementation of governance around their AI models and regulatory technology ( regtech ) tools should be able optimize their AFC programs , while at the same time safeguarding privacy rights and mitigating other risks linked to AI .
EU ’ s Artificial Intelligence Act and its applicability to AFC programs
In the EU ’ s Artificial Intelligence Act ( AI Act ), which entered into force on August 1 , 2024 , the EU described a key characteristic of an AI system as by using machine learning the system has the capability to learn , reason and model . It can derive models and algorithms from inputs or data and can generate outputs such as predictions , content , recommendations or decisions that can influence physical and virtual environments . AI systems are designed to operate with varying levels of autonomy depending on the AI system ’ s self-learning capabilities that allow it to change while in use , according to Article 12 of the AI Act . 4
Like with the anti-money laundering / counter-terrorist financing ( AML / CTF ) regime , the core principles of the EU AI Act are based on the risk-based approach .
Annex III of the EU AI Act assigns different risk categories to AI systems ( see Graphic 1 below ). High-risk AI systems include “ those intended to be used by or on behalf of law enforcement authorities , or by Union institutions , bodies , offices or agencies in support of law enforcement authorities or on their behalf to assess the risk of a natural person becoming the victim of criminal offences ,” e . g ., those that could form part of AML / CTF compliance programs . 5
AI systems used for AML / CTF would be considered high-risk systems and therefore subject to the following obligations prior to being allowed to be sold on the market . 6 The regulation of high-risk AI systems will take effect between 2026 and 2027 . 7
The risk areas are set out in Graphic 1 below . 8
Graphic 1 : AI system risk categories according the EU AI Act
NO RISK LOW RISK
HIGH RISK
AI Act allows the free use of minimal-risk AI . This includes applications such as AI-enabled video games or spam filters
AI systems with specific transparency obligations such as chatbots with the obligation to notify humans that they are interacting with an AI system
AI systems identified as high-risk include :
▪ critical infrastructures
▪ educational or vocational training
▪ safety components of products
▪ employment , management of workers and access to self-employment
▪ essential private and public services
▪ law enforcement that may interfere with people ’ s fundamental rights
▪ migration , asylum and border control management
▪ administration of justice and democratic processes
UNACCEPTABLE RISK
AI systems considered a clear threat to the safety , livelihoods and rights of people are banned , from social scoring by government to toys using voice assistance that encourage dangerous behavior .
Source : The AI Act 9 and the European Commission 10 ; Visualization by : Jennifer Hanley-Giersch
ACAMS Today | March – May 2025 51