Technology is constantly evolving, but one thing that has always fascinated people is the concept of artificial intelligence (AI) (B.Y. Kassa and E.K. Worku, “The Impact of Artificial Intelligence on Organizational Performance: The Mediating Role of Employee Productivity,” Journal of Open Innovation: Technology, Market, and Complexity, vol. 11, no. 1, 2025, p. 100474, https://tinyurl.com/5n8uzk8u; P. Radanliev, “Artificial Intelligence: Reflecting on the Past and Looking Towards the Next Paradigm Shift,” Journal of Experimental & Theoretical Artificial Intelligence, 2024, pp. 1-18, https://tinyurl.com/4tx7xjc5). AI systems undergo training to learn data patterns, make predictions, and carry out tasks typically associated with human intelligence (X. Gao and H. Feng, “AI-Driven Productivity Gains: AI and Firm Productivity,” Sustainability, vol. 15, no. 11, 2023, p. 8934). Generative AI uses deep learning models (i.e., computer systems trained on large amounts of data) to create human-like content in response to complex and varied prompts (J.C. Luna, “The Top 10 ChatGPT Alternatives You Can Try Today,” Datacamp, 2024, https://tinyurl.com/ycxw5638). One type of generative AI, ChatGPT 3.5, is a free version of an OpenAI nonprofit research laboratory and was launched in 2022. ChatGPT 3.5 is a platform capable of engaging in and refining natural language conversations toward a desired length, level of detail, style, and language (B. Marr, “A Short History of ChatGPT: How We Got to Where We Are Today,” Forbes, 2023, https://tinyurl.com/2pmzdrrr).
ChatGPT is one of many generative AI tools that have emerged in recent years. Comparable tools include Google’s Gemini and Microsoft Copilot (Luna 2024). While Gemini excels in bidirectional training for language understanding, Copilot allows users to search for specific information, generate text like emails and summaries, and create images based on written text prompts. Recent studies highlight ChatGPT’s superior performance in creating human-like text thanks to its vast training data and transformer-based architecture (Ö. Aydin and E. Karaarslan, “Is ChatGPT Leading Generative AI? What Is Beyond Expectations?” Academic Platform Journal of Engineering and Smart Systems, vol. 11, no. 3, 2023, pp. 118–134, doi:10.21541/apjess.1293702; M.N. Chu, “Assessing the Benefits of ChatGPT for Business: An Empirical Study on Organizational Performance,” IEEE Access, vol.11, no.1, 2023, pp. 76427–76436, doi:10.1109/ACCESS.2023.3297447).
Given these unprecedented changes in technology, auditors are constantly on the lookout to implement new tools to improve the efficiency of the financial audit process (R. Pratt and A.R. Otero, “Assessing the Integration of ChatGPT in IT Audits that Support Financial Statement Audits,” International Journal of Advanced Information Technology, vol. 14, no. 4, 2024, pp. 1–15. DOI: 10.5121/ijait.2024.14401). Financial statement audits can be performed by external auditors (EA) or internal auditors (IA). Both EA and IA acknowledge the significance of adopting innovative solutions like AI to improve the efficiency of their audits (G. Almufadda and N.A. Almezeini, “Artificial Intelligence Applications in the Auditing Profession: A Literature Review,” Journal of Emerging Technologies in Accounting, vo. 19, no. 2, 2022, pp. 29–42; A.R. Otero, Information Technology Control and Audit, 5th ed., CRC Press and Auerbach Publications, 2018, pp. 201–236).
Advantages of ChatGPT in Financial Audits
Data analytics and processing. In financial audits, the volume of data to be analyzed can be overwhelming, with traditional tools such as ACL and IDEA offering structured querying capabilities but often proving time-consuming and susceptible to errors (A. Fedyk, J. Hodson, N. Khimich, and T. Fedyk, “Is AI Improving the Audit Process?” Review of Accounting Studies, vol. 27, no. 1, 2022, pp. 938–985). Although these tools have supported auditors for decades, emerging technologies like ChatGPT provide a distinctive advantage through natural language processing, enabling auditors to interact with data more intuitively. The reliability of ChatGPT’s outputs is heavily dependent on data quality; thus, it is essential to have clean data to ensure accuracy and trustworthiness, particularly in contexts where decision-making hinges on dependable information (D. Leocádio, L. Malheiro, and J. Reis, “Artificial Intelligence in Auditing: A Conceptual Framework for Auditing Practices,” Administrative Sciences, vol. 14, no. 10, 2024, p. 238, https://tinyurl.com/4xbrzmdp). When applied to cleaned datasets, ChatGPT allows both EA and IA to efficiently process large volumes of information, extract key insights, and detect risk patterns through conversational prompts, surpassing the analytic depth of traditional tools (Almufadda and Almezeini 2022). This capability is particularly valuable in audit procedures like risk assessments and the testing of automated controls.
Continuous learning and adaptability. Continuous learning, when related to AI, can revolutionize the financial audit process by enabling machine learning algorithms to evolve and enhance auditors’ understanding of complex financial data patterns over time (M.M. Dong, T.C. Stratopoulos, and V.X. Wang, “A Scoping Review of ChatGPT Research in Accounting and Finance,” International Journal of Accounting Information Systems, vol. 55, no. 1, 2024, p. 100715, https://tinyurl.com/4pcwfue6). By enabling AI to cleanse data prior to analysis, a significant advantage can be achieved, particularly through continuous learning, which allows ChatGPT to detect subtle anomalies, outliers, and nuanced trends that, although potentially identifiable through regression analysis, may still be missed by human auditors (Dong et al. 2024). Furthermore, ChatGPT could learn to recognize subtle indicators of revenue manipulation by continuously analyzing transaction patterns, seasonal trends, and industry benchmarks.
The adaptive capabilities of ChatGPT make it a valuable asset for organizations navigating ever-changing regulatory requirements (B. Hacker, “Will ChatGPT Revolutionize Accounting? The Benefits of AI in Accounting,” ECONSTOR, 2023, https://tinyurl.com/3zk7ecvc). ChatGPT can quickly adapt and incorporate changes in regulations into its understanding, ensuring alignment with the latest audit regulatory requirements. By leveraging ChatGPT, organizations can streamline their compliance processes and effectively incorporate new regulations into their financial auditing practices (Dong et al. 2024; Pratt and Otero 2024).
Predictive analytics. Predictive analytics allow EA and IA to validate historical financial information and obtain an understanding of the potential trajectory of an organization’s financial performance. Forecasting financial trends, for example, is a prime application of predictive analytics in internal audits (M. Eulerich and D.A. Wood, “A Demonstration of How ChatGPT Can Be Used in the Internal Auditing Process,” 2023, https://tinyurl.com/5a6jzjx8). While traditional techniques such as regression and correlation analysis have long enabled auditors to identify patterns in historical financial data, ChatGPT offers a more accessible and scalable means of uncovering complex or non-linear patterns that may otherwise be overlooked or underutilized (Fedyk et al. 2022). By projecting these patterns into the future, auditors can gain a better understanding of an organization’s financial performance.
Audit efficiencies. When integrated into structured audit workflows, ChatGPT can enhance efficiency for both EA and IA by supporting routine tasks such as data extraction, preliminary data cleaning, analysis, and the review of key documents, such as financial statements, audit working papers, and internal control records (Almufadda and Almezeini 2022). While the initial implementation of ChatGPT in a first-year financial audit may introduce additional time and complexity, its integration in subsequent audit cycles can equip auditors with enhanced capabilities to improve existing procedures, such as generating faster, higher-quality work papers and performing consistent, automated testing of financial transactions (Fedyk et al. 2022). ChatGPT can also quickly analyze significant amounts of financial data, identify patterns, and flag discrepancies or anomalies that require further investigation (Dong et al. 2024).
Cost savings. ChatGPT provides cost savings by automating workflows and tasks, such as data entry, reconciliation, and basic analyses (S.A. Emett, M. Eulerich, E. Lipinski, N. Prien, and D.A. Wood, “Leveraging ChatGPT for Enhancing the Internal Audit Process–A Real-World Example from a Large Multinational Company,” 2023, https://tinyurl.com/mup9mhfr). In addition, ChatGPT can streamline the identification of discrepancies and trends, decreasing the time and resources spent on manual reviews (Eulerich and Wood 2023).
Disadvantages of ChatGPT in Financial Audits
Data security and privacy concerns. Integrating ChatGPT into financial audits raises significant concerns about data security and privacy (H. Alshurafat, “The Usefulness and Challenges of Chatbots for Accounting Professionals: Application on ChatGPT,” 2023, https://tinyurl.com/mubjcych). A major security concern stems from the cloud-based nature of ChatGPT. Although external auditors have their own policies prohibiting client data to be stored on the cloud, these policies are not always followed (A.J.G. Silvius and T. Dols, “Factors Influencing Non-Compliance Behavior Towards Information Security Policies,” CONF-IRM 2012 Proceedings, 2012, p. 39, https://tinyurl.com/wfs634cc). The above may result in private data being stored on third-party servers, exposing sensitive financial information and potentially causing financial harm. ChatGPT could also inadvertently expose private data to unauthorized parties, especially if security measures are not implemented.
Training data bias. ChatGPT’s training data may contain inherent biases, which can inadvertently reinforce existing prejudices or disseminate misinformation (D. Kalla and N. Smith, “Study and Analysis of ChatGPT and its Impact on Different Fields of Study,” International Journal of Innovative Science and Research Technology, vol. 8, no. 3, 2023, pp. 827-833). To mitigate these risks and promote fair, unbiased interactions, careful curation of training data and continuous monitoring of model behavior are essential (Dong et al. 2024). Within the context of financial statement audits, addressing training data bias poses significant challenges due to the limited volume of client-specific data and the technical difficulties associated with aggregating and standardizing data from multiple audit clients (each of whom may utilize distinct accounting systems and customized general ledger structures). Consequently, the development of diverse and representative datasets demands not only careful selection, but also robust normalization processes to ensure structural consistency and comparability. Both EA and IA should undertake comprehensive evaluations to identify and mitigate any underlying biases in the data used to train and fine-tune ChatGPT.
To mitigate training data bias risks and promote fair interactions, careful curation of training data and continuous monitoring of model behavior are essential.
Limited understanding of regulations. The landscape of financial regulations is dynamic, featuring frequent updates and changes. Consequently, integrating ChatGPT into financial audits poses a challenge in ensuring that the platform stays abreast of the latest regulatory developments (Dong et al. 2024). Unlike human auditors, ChatGPT lacks the inherent ability to track and understand evolving financial regulations. This limitation can lead to compliance risks, as external and internal auditors might rely on outdated or incomplete generated information when utilizing ChatGPT for their regulatory compliance assessments.
Inability to verify source data. Auditors rely on the accuracy and reliability of the data used to draw meaningful conclusions. Yet ChatGPT cannot verify or ensure the accuracy or integrity of source information, posing a notable challenge to its use in financial audits. The inability to cross-check and validate original sources in an audit could result in the acceptance of erroneous financial data or the overlooking of critical discrepancies (Emett et al. 2023). This lack of verification undermines the audit’s integrity, causing auditors to base their decisions on incomplete or misleading data, thereby increasing the risk of financial misstatements and compliance issues.
Explore and Adapt
The integration of ChatGPT into financial audits represents a notable development in the evolution of traditional auditing practices, with early academic studies highlighting its potential to enhance audit efficiency, regulatory compliance, and data analysis capabilities (Dong et al. 2024; Leocádio et al. 2024; Otero 2024). Successfully using ChatGPT in financial audits depends on a close partnership between human auditors and AI as they work together in this intricate process.
Despite the disadvantages documented above, the efficiencies and time-saving gains from utilizing ChatGPT in financial audits may encourage select segments of the profession, particularly larger CPA firms and those auditors serving complex or high-volume clients, to increasingly explore and adopt this technology in the execution of their audit procedures.
