Artificial Intelligence – Getting the AI Act Ready for Business

The EU’s AI Act was adopted in July 2024 and it will fully start to apply on 2 August 2026. To support the AI Act application, the European Commission published Guidelines on the definition of an artificial intelligence system and is preparing its Guidelines on the classification of AI systems as high-risk, which should be ready by February 2026. These guidelines should be also accompanied with a comprehensive list of practical examples of use cases of AI systems that are high-risk and not high-risk.

AI Act sets out that AI systems referred to in Annex III shall be considered to be high-risk. Annex III includes AI systems intended to be used to evaluate the creditworthiness of natural persons or establish their credit score, with the exception of AI systems used for the purpose of detecting financial fraud.

As implementation begins, it becomes apparent that several key issues remain unresolved, and need clarification to ensure the AI Act is workable in practice. The financial services sector faces important challenges in interpreting and implementing its provisions. For the leasing and rental industry, it is essential to make a clear distinction between adaptive AI systems and traditional statistical models. While the Commission’s guidance on the AI definition provides some reassurance about this, it can be still open to interpretation whether long-established statistical models are included in the AI definition. Leaseurope, together with other financial services stakeholders, engaged with policymakers and successfully contributed to shifting the Commission’s position toward a narrower interpretation of the definition.

Leaseurope calls for a clear distinction between genuine AI systems, i.e. those that exhibit autonomy, adaptiveness, and inferential reasoning, and traditional rule-based models such as logistic regression. The later have been used in credit scoring for a long time. Models like logistic regression are deterministic, interpretable, and do not evolve post-deployment. Including such models in the scope of the AI Act would derive from the Act’s risk-based approach, hinder supervisory efforts, and create unnecessary burden for businesses using transparent and compliant methodologies. Furthermore, to ensure a proportionate AI framework, it is important to consider as high-risk the AI used in the actual creditworthiness assessment step, i.e. the evaluation of repayment capacity, not AI systems used elsewhere, such as in customer service, fraud prevention, or marketing.

At the same time, the European Parliament’s Committee on Economic and Monetary Affairs (ECON) is preparing its own-initiative report on AI in financial services, expected for adoption by the end of 2025. The rapporteur’s draft report puts the focus on enabling innovation, rather than hindering the use of AI in financial services. The draft report concludes that further sectoral legislation to regulate AI in financial services is not necessary and calls for consistent definitions and the simplification of the framework to avoid duplications. On the other hand, the amendments tabled to the text focus on possible risks arising from AI, inherent biases and possible financial stability risks. The report should be adopted by the end of 2025.

Leaseurope continues to engage with stakeholders to ensure that our messages are understood and that compliance with AI Act can happen in a workable and balance manner.