Purpose: In addition to benefits, the adoption of artificial intelligence technology in banks is associated with costs and risks. The fragmented nature of prior research findings highlights the need for a comprehensive understanding of the implications of AI adoption in the banking industry in order to support the development of more effective implementation processes. Accordingly, this study aims to examine the implications of AI adoption in the banking industry.

Methodology: In terms of purpose, the study is applied, and in terms of data, it adopts a qualitative approach. Relevant sources were identified through a systematic search using Persian and English keywords across major national and international databases, including Scopus, Web of Science, ScienceDirect, Noormags, Magiran, and others, covering the period from 2010 to 2026 and from 1389 to 1405 in the Iranian calendar. Following the screening process, 32 studies were selected for final analysis. Data were analyzed using Sandelowski and Barroso’s seven-step meta-synthesis method. The credibility of the classifications and findings was assessed through Cohen’s kappa coefficient, face validity, and expert judgment.

Findings: The analysis resulted in the extraction of 169 basic codes, which were subsequently compared, consolidated, and classified into 65 conceptual codes and 13 main categories. The identified categories included economic, operational and process-related, strategic and organizational benefits, among others; explicit and hidden costs; and technical, data, security, governance, implementation, and other risks.

Originality/Value: The results indicate that AI adoption in the banking industry is a multidimensional phenomenon that cannot be evaluated solely in terms of efficiency gains or cost reduction. The proposed framework can assist bank managers, policymakers, and regulatory authorities in making more accurate and balanced decisions regarding the design, implementation, and development of AI systems by simultaneously assessing their benefits, explicit and hidden costs, and implementation risks.