Research

2.5 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Silvia Allegrezza (PI)
Dirk Zetzsche, Elise Poillot, Martin Stierle, Gabriele Lenzini (Co-PIs)
Eirini Botza (Partner)

The AI-LakeLegal project aims at a legal and regulatory analysis of the datalake idea provided as part of the FinTech initiate framework. The analyse focuses on the impact of big data processing operated by automated systems implying the support of artificial intelligence. The basic idea of how interoperability of data lakes is influenced by the use of the AI technologies will be analysed according to the recent legislation approved by the European Union with the AI act. The fast adoption of AI technology left behind whoever aims to regulate it. AI expedites the technological developments in the world of finance and we need to reassess from different perspectives whether the legal framework is still fit for purpose or needs improvement. It also imposes a weighty responsibility on businesses to articulate complex financial information to consumers. AI makes this information – the data lakes – more accessible and serves as a vigilant watchdog, swiftly detecting lapses in regulation clarity and enforcement weaknesses.
The following pillars will be developed: financial law; intellectual property; consumer protection; financial investigation. The datalake raises substantial legal issues in at least four dimension: a) data protection and consumer rights, b) financial regulation relating to operational risk and business continuity management as well as outsourcing as well as private law matters of liability, c) AML/KYC and criminal law accountability, and use as well as d) property rights in data stored on a blockchain and intellectual property in the processes used for the datalake.
The project thus addresses a core mission of the Future Fintech: the datalake project, delivering groundbreaking research results with practical use at a core element of the The LakeLegal project aims at a legal and regulatory analysis of the datalake idea provided as part of the FT initiate framework. This is a legal-only project as the datalake raises substantial legal issues in at least four dimension: a) data protection and consumer rights (Poillot), b) financial regulation relating to operational risk and business continuity management as well as outsourcing as well as private law matters of liability (Zetzsche), c) AML/KYC and criminal law accountability and use (Allegrezza) as well as d) property rights in data stored on a blockchain and intellectual property in the processes used for the datalake (Stierle).

In addition, legal interoperability is necessary to practically enable and realise a secure and legal flow of data in and from the datalake, in accordance with the above legal and regulatory analysis. The current lack of interoperability has been identified by the EU Commission as a significant obstacle to a thriving digital economy, and only by identifying the legal requirements for (technical solutions of) data sharing across data lakes it will be possible to ensure that financial transactions, based on aggregated data, are lawful and compliant.

The project thus addresses a core mission of the Future Fintech: the datalake project. Given that datalakes as such have not yet been examined from a legal perspective, yet provide significant legal challenges, the project promises to deliver groundbreaking research results with practical use at a core element of the FutureFinTech topics.

2.7 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Domenico Bianculli (PI)
Dirk Zetzsche (Co-PI)
Sallam Abualhaija (SPOC)

Regulations evolve over time through the amendment, repeal, or addition of legal provisions. The finance sector is a highly regulated domain that has undergone continuous regulatory changes following the 2008 crisis. Software-based financial services, such as online banking and trading, must consistently adhere to these regulations.
Monitoring and analysing regulatory changes is essential to ensure that such services remain compliant. These changes can have a significant impact on existing software systems that were previously compliant. However, manually tracking regulatory changes is both time-consuming and prone to errors.

This project explores innovative methods for automating the impact analysis of changes in financial regulations. Our aim is to characterise regulatory changes relevant to financial regulations and develop automated solutions to identify, classify, and assess the impact of these changes on existing (potentially compliant) software systems.

2.7 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Dirk Zetzsche (PI)
Mike Papadakis (Co-PI)
Dr Marco Bodellini (SPOC)

The AFS project explores the potential of automating the fund set-up and registration process with the CSSF. By streamlining these procedures, the project aims to enhance efficiency and ensure seamless data utilisation in subsequent supervisory processes.

As an outcome, we will identify and propose potential technologies that can be leveraged for key regulatory steps, paving the way for future practical applications. This initiative aligns with the FutureFinTech objective of automating financial services and addresses a fundamental aspect of the supervisory process for investment funds.

2.8 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Denitsa Stefanova (PI)
Radu State (Co-PI)
Denitsa Stefanova (SPOC)

With the widespread digitalisation of socio-economic interactions and financial transactions, data is emerging as a valuable asset class. Individuals must develop an awareness of their responsibility when sharing personal data and understand the long-term implications. However, this presents a significant challenge, as many are accustomed to delegating data-related decisions to third parties.
The two primary objectives of D@A are:

1. To understand and analyse data as an asset class.
2. To develop a decentralised digital market for tokenising and trading data.

These objectives highlight the interdisciplinary nature of our proposal and establish a strong connection to the FutureFinTech goals.

2.7 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Mats Brorsson (PI)
Denitsa Stefanova (Co-PI)
Mats Brorsson (SPOC)

Data science modelling and analysis using machine learning remains largely confined to computer and data science specialists, who often lack the necessary domain expertise. This is particularly evident in financial institutions, which serve as custodians of vast datasets. Professionals in these institutions would need to invest significant time in acquiring such a diverse skill set.
This project applies design research principles in the context of credit risk assessment and aims to develop automated mechanisms that translate domain knowledge into specifications, guiding the creation of purpose-driven and practical machine learning algorithms.

2.7 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Domenico Bianculli (PI)
Stanislaw Tosza (Co-PI)
Nicolas Sannier (SPOC)

Financial regulators ensure that investment fund documents comply with the regulations set out in national and international laws. These regulations specify the key information and metadata that must be included in each type of document.
Currently, compliance checking remains a manual process, which is both time-consuming and prone to errors. Moreover, compliance checking is inherently incremental, requiring regulators to carefully review multiple intermediate draft versions of a financial document before a fund is approved for listing on the financial market.
This project explores how to automate compliance checking as an incremental process. Specifically, it aims to develop innovative compliance and document engineering techniques—drawing inspiration from software engineering methods—to process document differences and generate corresponding incremental compliance checks.

2.9 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Stanislaw Tosza (PI)
Djamila Aouda, Raphael Frank (Co-PIs)
Olivier Voordeckers (SPOC)

The European Union’s anti-money laundering (AML) policy imposes several compliance obligations on financial institutions, particularly in relation to customer due diligence and reporting requirements. These obligations require institutions to assess the risk of money laundering associated with specific clients and transactions.
The primary objective of this project is to explore and contribute to the development of machine learning tools for AML compliance while also examining the legal implications of their use. This includes a critical analysis of the existing legal framework to ensure alignment with regulatory requirements.

2.7 years

The Government of the Grand-Duchy of Luxembourg – Ministry of Finance,
Luxembourg National Research Fund,
University of Luxembourg

Domenico Bianculli (PI)
Michael Halling (Co-PI)
Marcello Ceci (SPOC)

Monitoring fund activities, such as share subscriptions, is crucial for protecting investors’ rights and combating financial crime, including fraud and money laundering. However, manual monitoring is challenging and prone to errors, primarily due to the complexity of financial regulations, the intricacies of fund documents, and the vast amount of data involved.

To address these challenges, this project aims to develop an automated, scalable, and accurate approach for monitoring fund activities using Natural Language Processing, Machine Learning, and Runtime Verification. Specifically, the project will focus on two key objectives:
1) AI-driven extraction of metadata and intermediate semantic representations from financial regulations and fund documents.
2) Automated monitoring and verification of financial transaction records against regulatory requirements derived from these documents.