In our response, ACCIS notes that credit reference agencies are exploring the use of AI to help lenders evaluate the creditworthiness of individuals and businesses, as well as to establish credit scores. We highlight that AI’s ability to analyse vast and complex datasets enables it to identify patterns and correlations that traditional methods might overlook.
At the same time, we emphasise that traditional statistical techniques, particularly logistic regression, have long been the foundation of credit scoring. Even today, approximately 90% of credit scoring still relies on these simpler statistical methods. Logistic regression remains a key tool in credit scoring due to its simplicity, transparency, interpretability, and regulatory acceptance. These qualities make it easier for lenders and regulators to understand and trust the scoring process.
Therefore, we argue that logistic regression should not be classified as AI under the AI Act.