New Algorithmic Credit Scoring Models: What Protections for the Consumer? (Actualidad jurídica iberoamericana)

Summary

• Directive 2023/2225/UE regulates consumer creditworthiness evaluation under Article 18 and will apply from 20 November 2026.p.4
• Italy transposed Directive 2023/2225/UE through Legislative Decree 31 December 2025, no. 212.p.4
• The CJEU ruling of 11 January 2024 in Case C-755/22 confirmed that bank credit checks serve to prevent consumer over-indebtedness.p.6
• The credit scoring process is structured into 3 macro-phases: data warehousing, data mining, and data correlation.p.11
• The European AI Act entered into force on 2 August 2024 and general provisions apply from 2 August 2026.p.37

Key insights

1. Traditional credit scoring systems rely on hard data such as income, existing debts, and account history to assess borrower solvency.p.12
2. Modern credit scoring models utilize machine learning to combine hard data with alternative soft data extracted from digital footprints.p.13
3. Algorithmic credit scoring creates severe discrimination risks when machine learning algorithms reproduce historical biases.p.15
4. The black box architecture of self-learning algorithms hides internal mechanics, making algorithmic credit decisions hard to explain.p.16
5. Article 22 of the GDPR establishes a general prohibition on decisions based solely on automated processing that significantly affect individuals.p.22
6. Article 18(8) of Directive 2023/2225/UE grants consumers explicit rights to human intervention, clear explanation, and a full decision review.p.34
7. AI systems used to evaluate consumer creditworthiness are categorized as high-risk under the AI Act due to their impact on access to essential services.p.39