Developing an Audit Model Based on Autonomous Artificial Intelligence (Agentic AI): Opportunities, Risks, and Corporate Governance Requirements
Developing an Audit Model Based on Autonomous Artificial Intelligence (Agentic AI): Opportunities, Risks, and Corporate Governance Requirements
َAbolfazl Aminian1
1) Assistant Professor, Department of Accounting, Payame Noor University, Tehran, Iran
IOI Code
:
AAHC-ABHKZK
Publication :
National Conference on Management, Commercialization, Knowledge-Based Economy and Investment for Production(impeb.ir)
Abstract :
The development of agent-based artificial intelligence (Agentic AI) technologies, which are based on autonomous agents capable of planning, reasoning, and independently executing complex tasks, has fundamentally transformed the landscape of the audit profession. Unlike the previous generation of generative AI, which was mainly used as an auxiliary tool for preparing reports, producing documents, and summarizing information, intelligent agents are able to independently manage a significant part of the audit processes. These systems are capable of executing multi-step workflows, audit tests, continuous reconciliation of financial data, and evidence evaluation with minimal human intervention. In this regard, recent studies have introduced the concept of “agent-based auditing” as a new approach in which autonomous agents perform audit tasks in an integrated manner and within the framework of human supervision. Also, research evidence shows that large language models have high accuracy in verifying accounting information and reconciling financial data in real time; However, the results of some field studies indicate that the use of these technologies alone does not guarantee the improvement of the quality of professional judgment or the increase in adherence to audit guidelines, and their effectiveness depends largely on the experience and skill of the auditor. Along with these opportunities, the development of autonomous agents has also created new challenges in the field of corporate governance and assurance systems. The increase in the speed, volume and complexity of AI-based decisions has caused traditional audit methods to not meet all supervisory needs. In addition, the lack of comprehensive frameworks for authenticating intelligent agents, determining the boundaries of responsibility, traceability of decisions and algorithmic accountability are among the most important concerns of legislative institutions and the audit profession. In recent years, the development of governance frameworks such as the Singapore Agent-Based AI Governance Framework and the European Union Artificial Intelligence Regulations reflect the efforts of legislative systems to create effective mechanisms in the field of human oversight, transparency, security and accountability of these technologies. Using a conceptual-analytical approach and based on a systematic literature review, the present study seeks to present a model for employing autonomous AI agents in the audit process; a model that, while utilizing the capacities of this technology in increasing efficiency, continuous auditing, and complete coverage of financial transactions, also includes the requirements related to transparency, explainability, algorithmic risk control, human oversight, segregation of duties, and accountability. It is expected that the findings of this study can provide a suitable basis for developing corporate governance policies, revising audit standards, and designing supervisory mechanisms appropriate to AI-based environments.
Keywords : agent based artificial intelligence continuous auditing corporate governance large language models algorithmic accountability