What changed between the draft committee report and the plenary report

This report examines the use and impact of AI in the financial services sector and the regulatory landscape. The Rapporteur provides policy recommendations to enable the use of AI in financial services and clarify regulatory overlaps. The report addresses aspects specific to the financial services sector and does not cover matters falling within the remit of other Committees.

The Rapporteur believes that it is crucial for the policy debate on AI in financial services to be grounded in reality and focused on tangible and plausible questions. Due consideration must be taken of the existing legal framework and the practical realities of the technology’s use in financial services, rather than speculate about abstract or theoretical concerns.

The report therefore starts by analysing the deployment of AI in the sector. It notes that the majority of AI use cases aims to cut costs by streamlining operations, rather than create new revenue streams. Most use cases represent low-hanging fruit rather than high-risk innovation, meaning that it is safe to say that deployment of AI in finance has been prudent. We are far from experiencing a financial system run, or heavily dependent on, autonomous, auto-pilot AI models that threatens financial stability and consumers’ interests.

The reality is the opposite: the sector is so heavily regulated, and the fiduciary responsibility of financial institutions so highly regarded, that the lion’s share of use cases are both low-risk and include a human expert in the loop. Nonetheless, the diffusion and uptake of AI technologies across the financial services sector holds significant potential. Not only it can improve the sector’s efficiency, enhance consumer services, and strengthen the competitiveness of European firms, but it can also support more effective anti-money laundering and fraud detection.

That is not to say that AI deployment in financial services is without risks. The issue of data quality, explainability and transparency of AI is a challenge in this domain as within others. However, the financial services sector with its myriad of detailed directives and regulations, is well positioned to handle these risks. Financial institutions, whether it be banks, insurances undertakings or asset managers, are required by EU financial services legislation to have systems in place for data quality, data lineage, data governance, operational resilience, outsourcing, model risk, concentration risks, discriminatory outcomes, and more, which provides a framework for AI deployment and governance. As deployment of AI in finance continues, it will be critical to continue monitoring these risks and to provide finance experts with resources, training and AI-literacy.

The alternative is to take a restrictive approach to AI deployment in finance, with new legislation out of fear of the unknown effects, or because status quo is comfortable. Such a policy would deprive the financial services sector of the opportunity to use AI. This would ultimately undermine the sector’s competitiveness, the quality of services offered, and the benefits delivered to consumers. It would also have a negative impact on investment in AI technologies, considering that the financial services sector is the biggest spender on ICT services and products. Such a route should be off the table considering the global race for AI, the stark geopolitical realities underpinning it, and the fact that the EU is already lagging behind.