THE IMPACTS OF ARTIFICIAL INTELLIGENCE ON FINANCIAL SERVICES

Seminar organized by the Association Europe-Finances-Régulations (January 29, 2025)

On the eve of the AI summit, the AEFR organized a seminar to answer questions about the “transformative potential” of AI in financial services: how can it change practices in customer relations (KYC), cyber-security and the fight against fraud and money laundering.

AI and France’s Competitiveness

Philippe Aghion (professor at the Collège de France) wrote a report in 2024 entitled “Our ambition for France”, the conclusions of which are « reasonably optimistic » for our country. Symbolic, numerical and generative AI, as well as LLMs, contribute overall to improving the efficiency of production processes, making companies more competitive, profitable and job-creating. Recent studies show that AI is helping to accelerate innovation and increase patent applications. If is well regulated, AI’s contribution to French growth would be around 1% per year for about ten years, before falling due to system obsolescence and business concentration. However, the impact on employment would be very uneven depending on the sector of activity, the companies, the types of employment and even the workstations. Only 10% of jobs would be completely or partially threatened in Western countries.  The latest mapping of employment risks shows the importance of developing research on production processes and, in particular, on their flexibility.

AI and stock market fraud

Corentin Masson (chief AI engineer at the AMF) underlines the interest of AI in the investigations conducted by the AMF on the practices of listed companies, particularly in terms of compliance with sustainability, the fight against fraud (insider trading, price manipulation, market abuse, etc.). The objective of the AMF is to secure financial transactions, which implies in particular securing databases (aligned with the European taxonomy). He recommends publishing non-financial reports in HTML format with XDRL tags, in order to facilitate comparisons between companies and sectors of activity, and to set usable standards.

AI and the fight against money laundering

Fabrice Desprez (CEO of Discal, a subsidiary of KBC group) estimates that money laundering worldwide is worth nearly $4 trillion, with only $2 billion detected and $200 million recovered. At KBC, exposure to money laundering risk is assessed on the basis of 150 criteria. Approximately 10% of customers present risks, which are detected based on the particularity of the transaction and the customer’s score. He considers that the development of cryptos creates an additional risk of money laundering, and that Regulation (EU) 2023/1114 (known as MICA), which establishes uniform rules for crypto-asset issuers, is insufficient. He also regrets the lack of exchanges between banks on this subject. He also denounced the excessive regulation of banking activities.

AI and Banking Fraud

Mathilde Clauser (director at Revolut) presented the services offered by her bank (50 million customers), whose priority objective is to ensure the security of the operations of its business and retail customers.  She distinguishes between unauthorized fraud and “authorized fraud.” The latter covers in particular the extortion of funds by identity theft (visual and/or sound)  thanks to social networks (in particular META and X). She demonstrated, through use cases, the contributions of AI in detecting these frauds. She estimates that more than 90% of fraud is detected by IA at Revolut. AI and compliance management.

AI and compliance

Frédéric Boulier (Chief Compliance Officer at Oracle) presents the latest developments in AI for compliance management. He revealed that nearly €250 billion is spent each year by banks to combat financial insecurity and verify the compliance of sustainability reporting. He admits that unsupervised learning by generative AI creates value because it improves KYC and the number of supervisors (or data scientists) responsible for controlling and making better use of the extra-financial reporting of large companies, and soon, of mid-caps and SMEs, as well as correcting the biases observed in the software used to exploit the data of the reports.

AI and financial rating

Vincent Gusdorf (Managing Director at Moody’s) analyzed the contribution of AI in the financial rating process of listed companies (1). He argues that the latest generation software (such as GPT 4o and Deepseek) has a “reasoning capacity” that helps accelerate their efficiency in detecting information gaps from financial security issuers. Despite these advances, he believes, however, that investment risks are still insufficiently measured by rating agencies.

Report written by J-J.Pluchart

(1)       read the chapter by J-J. Pluchart, “Towards an ESG rating of credit securities issuers”, in C.de Bossiseu and D Chesneau, Réussir le transition énergétique et écologique, Eds Eska, 2024.