Getting the AI Act ready for reality

The EU’s AI Act was adopted in July 2024, and has now entered its critical implementation phase. Driven in part by political momentum to finalise the text ahead of the 2024 European elections, the regulation was not only adopted at speed, but also seeking to regulate  developments in a rapidly evolving field.

As implementation begins, it is increasingly clear that several key issues remain unresolved, and must be clarified to ensure the framework is workable in practice. The financial services sector faces important challenges in interpreting and implementing its provisions. For the consumer credit industry, maintaining a clear and workable distinction between adaptive AI systems and traditional statistical models is essential.

With the staggered application of different provisions, several key dates and pieces of guidance will be crucial for consumer credit providers to follow closely. However, the feasibility of these dates is still subject to on-going discussions. In the end of June, the question of pausing the rules was raised in top-level discussions between the Member States. The responsible Commissioner,  Henna Virkkunen, has also stressed that the necessary guidance should be in place for application of the framework.

What is AI?

In the beginning of 2025, the European Commission published its first guidance on the definition of an AI system under the AI Act. While the guidance provides some reassurance, it does not fully resolve concerns about the potential inclusion of long-established statistical models.  Through sustained engagement with policymakers, the Federation successfully contributed to shifting the Commission’s position toward a narrower interpretation; despite strong resistance from the drafters of the guidelines.

Eurofinas has continued to repeatedly emphasise the importance of drawing 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 decades. They are deterministic, interpretable, and do not evolve post-deployment. Including such models in the scope of the AI Act would run counter to the regulation’s risk-based approach, misdirect supervisory efforts, and create unnecessary burdens for lenders using transparent and compliant methodologies. Ensuring they are explicitly excluded is essential to preserving a practical and rational application of the legislation, which is the key objective of Eurofinas´ outreach.

What is part of high-risk operations?

Under the AI Act, AI systems used to evaluate the creditworthiness of natural persons will be classified as high-risk. However, there is still both significant ambiguity as well as and diverging views, on which elements of the credit lifecycle this classification is intended to cover.

To ensure the framework remains proportionate and targeted, Eurofinas strongly advocates for limiting the high-risk classification strictly to the creditworthiness assessment step, i.e. the evaluation of repayment capacity. AI systems used elsewhere in the lending journey,  such as for customer service, fraud prevention, or marketing. should not be swept into the high-risk category with the carried implications. This would expand the high-risk scope inappropriately, resulting in regulatory overreach and imposing disproportionate compliance burdens on providers.

A Commission consultation is currently on-going on the various high-risk AI operations, and it is open for responses until 18 July.

Gen AI guidance and political developments

The Commission has also conducted a consultation on general-purpose and generative AI. The consultation will inform guidance on how responsibilities are allocated between developers and deployers of such systems.

While consumer credit providers are generally not developers of generative AI models, their growing use in customer support and internal operations makes this an important area to watch.

In parallel, the European Parliament’s Committee on Economic and Monetary Affairs (ECON) is busy preparing its own own-initiative report on AI in financial services, expected for adoption by the end of 2025. The lead rapporteur, MEP Arba Kokalari (EPP, Sweden), also a Member of the Parliament´s AI Act implementation working group, wants to send a clear message in support of responsible innovation and against further legislative initiatives. She wants the focus to be on enabling, rather than hindering the use of AI in financial services.

Separately, the European Commission has confirmed that the AI Act will be reviewed as part of a broader fitness check of the EU digital acquis, scheduled for the fourth quarter of 2025 (see separate article). While the AI Act will be included in this review, the Commission has indicated that only limited adjustments should possibly be expected, primarily aimed at facilitating compliance for SMEs and easing the Act’s application without altering its core structure.

Eurofinas is actively engaging with both legislative and regulatory stakeholders to ensure that the needs of consumer credit providers are understood and that implementation remains balanced and workable.