BGL BNP Paribas pioneers AI governance in Luxembourg

BGL BNP Paribas is more than halfway through its three-year AI "metamodel" project, after in 2024 it became the first bank in the country to secure public co-financing for an AI project under Luxembourg's R&D and innovation aid scheme. With a total budget of approximately €7.4 million, the project is co-financed by the Ministry of the Economy up to €2.95 million. The objective is to develop a metamodel – an additional AI layer designed to monitor, evaluate and supervise other AI models in production.

Creating an independent, transparent and auditable system is essential in a highly regulated environment like finance.
Anne Goujon, BGL BNP Paribas

 “Creating an independent, transparent and auditable system is essential in a highly regulated environment like finance,” explains Anne Goujon, Director of Data Science Lab at BGL BNP Paribas. “The tool will primarily help us identify performance drift, emerging biases or unintended discriminatory behaviours, and early signals of deviation related to operational anomalies or data distortion.”

Building an AI governance framework for banking

The R&D project is conducted in partnership with the University of Luxembourg’s Interdisciplinary Centre for Security, Reliability and Trust (SnT). The SnT brings cutting-edge scientific expertise in AI, while the bank’s teams bring business knowledge, insights into regulatory constraints and real-world use cases. Together, the two partners work to develop a tool that has a real, positive impact on daily operations.

We are building an AI trust framework for the future.
Anne Goujon, BGL BNP Paribas

“Monitoring AI solutions in a fast-changing financial, regulatory and economic setting is very time-consuming and takes about 80% of our lab’s capacity,” says Ms Goujon. “This is why we are creating this intelligent assistant to give our data scientists a continuous and exhaustive view of the models in production, while still being able to focus on innovation, development and the integration of new AI components.”

She points out that many companies prioritise visible, high-impact AI use cases with a quick return on investment, while governance can be perceived as more complex and costly. BGL BNP Paribas chose the opposite route, positioning governance as a strategic foundation for scalability and alignment with EU regulations. “We are building an AI trust framework for the future.”

Luxembourg as a hub for financial sector AI innovation

As the first bank to secure public R&D funding for an AI project in Luxembourg, BGL BNP Paribas is taking a pioneering step in the financial sector, which is known for its cautious approach. “We see this as a strong signal to Luxembourg’s broader financial ecosystem,” says Ms Goujon.

Locating this R&D project in Luxembourg was a strategic choice for the bank. “Luxembourg has a highly international financial ecosystem, strong public support for innovation and easy access to high-level academic partners such as the SnT,” she points out. The project funding has also allowed both BGL BNP Paribas and the SnT to attract highly skilled professionals to the country.

Luxembourg has a highly international financial ecosystem, strong public support for innovation and easy access to high-level academic partners.
Anne Goujon, BGL BNP Paribas

Looking five years ahead, Ms Goujon expects BGL BNP Paribas to have increasingly self-monitored, transparent and explainable AI systems that can offer standardised real-time supervision across Europe. She hopes to position the bank as a benchmark in responsible AI with the capacity to scale and replicate the framework developed in Luxembourg across the BNP Paribas Group. “The strong collaboration between banks, regulators and research centres is a key competitive advantage that could contribute to positioning Luxembourg as a digital governance hub for financial institutions.”

Photo credits: BGL BNP Paribas

This article was originally published in Crossroads Magazine no. 9. Read the full edition below: