Auditing AI systems: Integrating ESG principles for sustainable and ethical AI deployment in the EU

The European Union Artificial Intelligence Act establishes a framework to ensure that artificial intelligence systems deployed within the European Union are safe, compliant with fundamental rights, and aligned with Environmental, Social, and Governance (ESG) principles. However, it lacks specific guidelines to address ESG trade-offs, such as balancing transparency with fairness or environmental concerns with societal impact. This article introduces a novel artificial intelligence audit framework that integrates ESG criteria using the Fuzzy Analytic Hierarchical Approach and Multicriteria Optimization. The framework acts as an institutional translation mechanism and enables decision-makers to prioritize ESG trade-offs and optimally allocate limited resources. By systematically identifying, prioritizing, and addressing the risks of the artificial intelligence system, the framework enhances innovation governance and ensures sustainable and ethical deployment of artificial intelligence. Applying this framework results in improved audit quality, enhanced risk mitigation, and continuous improvement of artificial intelligence systems, thereby promoting positive societal and environmental outcomes. This integrated framework bridges the gap between regulatory expectations and practical implementation, fostering responsible innovation in artificial intelligence in the European Union. Overall, we contribute to the extant literature by repositioning AI governance from a compliance-oriented activity to a dynamic capability that enables firms to manage ESG trade-offs and sustain responsible AI innovation under regulatory uncertainty.