AI Act: Simplification agreed and implementation continues

The implementation of the AI Act and development of connected measure are continuing to evolve as the EU institutions work to provide greater legal certainty ahead of key compliance deadlines. The European Parliament and the Council recently concluded the agreement on the AI Act simplification package under the Commission's Digital Omnibus initiative. Most notably, several implementation deadlines have been postponed, with obligations for high-risk AI systems now expected to apply from the end of 2027 instead of August 2026. 

Alongside the legislative changes, the Commission continues to develop implementation guidance. A public consultation on the draft guidelines for the classification of high-risk AI systems remains open until later this month, with final guidance expected before year-end. 

For the consumer finance sector, one of the most important outstanding questions remains the scope of the AI Act itself. While the final legislative text deliberately removed references to long-established statistical models, and the Commission's own guidance on the definition of an AI system points towards their exclusion, uncertainty persists over the status of models such as logistic regression. The Commission’s AI Office advocates for a broader interpretation, despite concerns raised by a majority of Member States and by institutions including the European Central Bank and the European Insurance and Occupational Pensions Authority. 

Eurofinas continues to engage with the AI Office, other Commission services and stakeholders across the financial services sector to promote proportionality and legal certainty. Together with Leaseurope and industry partners, Eurofinas advocates that logistic regression and comparable statistical models should be explicitly confirmed to fall outside the scope of the AI Act, while promoting coherence between the AI Act, the Consumer Credit Directive and existing data protection legislation.