Artificial Intelligence and Financial Reporting: A Conceptual Analysis of Accuracy, Transparency, and Decision-Making in Contemporary Business Organizations

Abstract

Financial reporting exists to convert an organization’s economic activity into information that external parties can trust and act upon, and artificial intelligence (AI) is now implicated on both sides of that exchange — assisting firms in the preparation of disclosures and assisting investors in interpreting them. This paper offers a conceptual analysis of the implications of that dual involvement for the three properties financial reporting is designed to deliver: accuracy, transparency, and support for sound decision-making. Drawing on recent accounting and finance research, including large-sample field evidence on generative AI use within actual reporting workflows, the paper argues that AI's effect on these three properties is not uniformly positive. Machine learning models have demonstrably improved specific accounting estimates and fraud-detection tasks, and generative AI is measurably present in firms' disclosure writing; at the same time, the same body of evidence documents emergent failure modes — systematic overreliance on incorrect AI-generated classifications, generative systems capable of producing fluent yet ungrounded disclosure text, and analytical tools whose reasoning remains difficult for even sophisticated users to peruse. The paper develops a conceptual framework in which accuracy and transparency function as preconditions for AI to improve decision-making, rather than as automatic corollaries of AI adoption, and traces the resulting implications for preparers, auditors, investors, and standard-setters. Its contribution is a structured account of where AI is presently strengthening financial reporting, where it is introducing new and consequential risks, and what conditions determine which outcome prevails in a given organizational context.