AI-Intensive Firms and ESG Performance: Evidence on Governance, Risk Management and Sustainable Value Creation

This study investigates whether firms that intensively adopt artificial intelligence (AI) exhibit superior sustainability outcomes compared to comparable non-AI-intensive firms. Using an observational design based on Propensity Score Matching (PSM), the analysis isolates the average treatment effect of AI intensity on multiple dimensions of ESG performance, including environmental, social, governance indicators, unmanaged ESG risk, and disclosure transparency. The sample comprises publicly listed firms in North America and Europe over the 2016–2024 period, with AI intensity classified through a rigorous two-stage procedure combining external stock lists, documentary evidence, patent-based signals, and descriptive analytics. ESG outcomes are drawn from independent providers (LSEG/Refinitiv, MSCI, Sustainalytics, Bloomberg) to minimize construct contamination and ensure robustness across methodologies. Results indicate that AI-intensive firms achieve significantly higher ESG performance, particularly in the governance dimension while also demonstrating superior transparency in sustainability disclosure. Bootstrap confidence intervals and Rosenbaum sensitivity bounds confirm the robustness of these effects, suggesting resilience to unobserved confounding within plausible bounds. These findings support the hypothesis that AI functions as an enabling mechanism for improved corporate governance, risk management, and responsible resource allocation.