The impact of AI-powered embedded finance platforms on credit risk management: new paradigms, opportunities, and challenges

Purpose- The purpose of this study is to examine, from a multidimensional perspective, the structural transformation created by artificial intelligence-supported embedded finance platforms in credit risk management. The invisible integration of financial services into digital ecosystems transforms credit risk assessment processes by moving them away from traditional banking towards a more dynamic, behavioural and data-based structure. Within this framework, the study aims to evaluate the effects of artificial intelligence algorithms on risk prediction performance, forms of data use and levels of transparency by analysing the platforms Stripe, Unit and Weavr in a comparative manner. Methodology- Adopting a qualitatively oriented exploratory approach, it examines multiple data sources consisting of platform documentations, financial technology reports and the academic literature through content analysis. In addition, an original conceptual model explaining artificial intelligence-based credit risk management in the context of embedded finance has been developed. This methodological approach makes it possible to evaluate both the technological infrastructure and the ethical and regulatory dimensions in a holistic manner. Findings- The study shows that the analysed platforms exhibit clear differences in data processing capacity, algorithmic modelling techniques and risk management approaches. While Stripe’s processing of behavioural data with deep learning models produces dynamic and highly accurate risk predictions; Unit’s transparent analytical structure based on regulatory compliance strengthens auditability in the credit assessment process; and Weavr’s modular risk engine allows flexible, sector-based credit assessments. However, it has been determined that critical problems such as data privacy, algorithmic bias and model explainability persist in all platforms. Conclusion- This study reveals that artificial intelligence-supported embedded finance platforms go beyond technical innovation in credit risk management and create a new financial architecture, and it offers original contributions to the literature both at the level of comparative platform analysis and at the level of a conceptual framework explaining the data–impact relationship. The findings constitute an important basis for future research on algorithmic transparency, regulatory compliance and the ethical boundaries of the use of alternative data. Keywords: Embedded finance, artificial intelligence, credit risk management, fintech, algorithmic bias JEL Codes: G20, G24, O33