The EACB welcomes the opportunity to contribute to the European Commission’s targeted consultation on Artificial Intelligence (AI) in the financial sector.
As of now, there is no immediate concentration risk in the financial markets related to AI. Financial institutions are utilising a range of third-party AI providers and continue to rely on in-house development and internal data assets, which mitigates the risk of over-dependence on any single external provider. Despite some trends towards herding around general-purpose AI and productivity tools, the presence of numerous niche and specialised AI providers within the financial sector helps maintain a diverse landscape.
Looking ahead, the situation is more uncertain. The impact of the AI Act remains to be seen, particularly regarding its effectiveness in building consumer trust versus its role as a supportive framework for AI development. There is a concern that Big Tech companies might dominate the AI market, leading to increased dependency and reduced competition.
In terms of reporting, AI shows promise in easing data collection and preliminary reporting tasks. It can automate data gathering from unstructured documents and generate initial reports, but full reporting processes still require human oversight due to their complexity. The AI Act introduces additional compliance requirements, which could increase the regulatory burden on businesses.
Regarding data access, while external datasets are beneficial, they are not a prerequisite for developing AI applications. Existing EU regulations, including the Data Governance Act and FiDA, already support data sharing without necessitating new public policy measures. Excessive regulation could hinder innovation and increase investment requirements.
Looking forward, AI is expected to drive changes in the financial sector, particularly in areas such as fraud detection, risk management, and personalised services. These advancements should be seen as enhancements to human expertise rather than replacements. AI’s role in financial services will likely continue to grow, but it is crucial to balance its implementation with careful oversight to ensure it supports rather than undermines human decision-making.
With regard to the AI Act, it marks a significant step in regulating AI systems but requires careful interpretation. The financial sector uses various statistical methods that are essential for its operations, and it is important to recognise that these methods, while valuable, may not meet the criteria set out by the AI Act. Including such methods within the Act’s scope might impose unnecessary regulatory burdens without a proportional increase in risk. Ongoing dialogue with the European Commission and other stakeholders is critical for the practical implementation of the AI Act. Clear communication and understanding will help ensure that the Act addresses real concerns without stifling innovation or imposing undue constraints on the use of statistical techniques in finance. In May, the EACB and other financial services associations requested that the AI Office provide clearer guidelines on what constitutes an AI system, particularly regarding statistical techniques like logistic regression. This request, shared with DG FISMA, aims to offer valuable input for refining the AI Act’s implementation.
Article 96 of the AI Act tasks the Commission with developing guidelines on applying certain requirements and obligations, considering the relevant harmonised standards and common specifications. The AI Office is also responsible for providing practical guidelines for high-risk AI use cases in the financial sector. These efforts should be sufficient for ensuring compliance without adding unnecessary complexity.
Finally, the European Supervisory Authorities (ESAs) are already working on AI-related priorities. For instance, the European Banking Authority (EBA), the European Securities and Markets Authority (ESMA), and the European Insurance and Occupational Pensions Authority (EIOPA) have ongoing projects and plans for addressing AI in financial services. These authorities will monitor AI developments and provide guidance as needed, indicating that additional guidelines may not be necessary in the short and medium term. In the long run, an assessment of the AI Act and its impact on the financial sector may be necessary before introducing legislative or non-legislative measures.