Kevin C. Moffitt
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Kevin C. Moffitt has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Miklos A. Vasarhelyi
Rutgers, The State University of New Jersey
3 shared publications - Daniel E. O'LEary
Marshall School of Business, University of Southern California, Los Angeles, CA, USA
2 shared publications - Dan Sunderland
Northeastern University
1 shared publication - Danyang Wei
Durham University
1 shared publication - Huaxia Li
University of Michigan–Dearborn
1 shared publication - Marco Schreyer
Swiss Federal Audit Office, Switzerland and UC Berkeley, USA
1 shared publication
Latest Publications
- Open Access
- 11 September 2026
- Published by Elsevier in SSRN Electronic Journal
Auditing Artificial Intelligence: A Research Agenda for AI Assurance
Artificial intelligence (AI) is becoming embedded in organizational processes, raising a fundamental question for auditors: how can they provide credible independent assurance over systems whose risks and dependencies evolve across models, data, components, organizational processes, and deployment environments? As these systems become more capable, interconnected, and context-dependent, established assurance principles remain relevant but become more difficult to apply. This paper develops a research agenda for auditing AI. First, it examines how advances in AI capabilities, system architectures, and deployment settings change what must be assured, which risks matter, and where to draw engagement boundaries. Second, it synthesizes developments in policy, regulation, professional and standard-setting activities, and technical AI research to characterize the emerging AI assurance ecosystem. Third, building on the risk-based auditing process, it identifies seven interconnected research directions spanning engagement design, criteria, procedures, evidence, professional judgment, reporting, and the conditions needed to sustain credible assurance. Together, these contributions provide an organizing framework for research on a central challenge facing the auditing profession: how to develop and sustain credible assurance as AI systems and their organizational uses continue to evolve.
- 11 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAuditing Artificial Intelligence: A Research Agenda for AI Assurance
Artificial intelligence (AI) is becoming embedded in organizational processes, raising a fundamental question for auditors: how can they provide credible independent assurance over systems whose risks and dependencies evolve across models, data, components, organizational processes, and deployment environments? As these systems become more capable, interconnected, and context-dependent, established assurance principles remain relevant but become more difficult to apply. This paper develops a research agenda for auditing AI. First, it examines how advances in AI capabilities, system architectures, and deployment settings change what must be assured, which risks matter, and where to draw engagement boundaries. Second, it synthesizes developments in policy, regulation, professional and standard-setting activities, and technical AI research to characterize the emerging AI assurance ecosystem. Third, building on the risk-based auditing process, it identifies seven interconnected research directions spanning engagement design, criteria, procedures, evidence, professional judgment, reporting, and the conditions needed to sustain credible assurance. Together, these contributions provide an organizing framework for research on a central challenge facing the auditing profession: how to develop and sustain credible assurance as AI systems and their organizational uses continue to evolve.
- Open Access
- 4 September 2026
- Published by Wiley in Contemporary Accounting Research
Voluntary Disclosure Through the Prominence of Risk Factors in the 10‐K
Prior research finds that the textual content of Item 1A risk factor disclosures in 10‐K filings provides valuable information about firm risk. However, less is known about whether the ordering of these disclosures conveys useful information. We examine whether the relative prominence of individual risk factors within Item 1A reflects firms' exposure to the underlying risks and predicts future adverse outcomes. Focusing on credit and goodwill risk disclosures, we find that risk factor prominence is associated with proxies for the underlying risks and predicts credit rating downgrades, bankruptcy filings, and goodwill impairments. We further find that prominence is more informative during periods of high information uncertainty, when the benefits of risk disclosure are predicted to be greater. Overall, our findings suggest that risk factor prominence offers a valuable signal of firm risk that complements the textual disclosures in Item 1A. Accordingly, investors, analysts, lenders, auditors, boards, and regulators should consider both the level of, and changes in, risk factor prominence when evaluating firm risk.
- 4 September 2026
- Published by Wiley in Contemporary Accounting Research
Open AccessVoluntary Disclosure Through the Prominence of Risk Factors in the 10‐K
Prior research finds that the textual content of Item 1A risk factor disclosures in 10‐K filings provides valuable information about firm risk. However, less is known about whether the ordering of these disclosures conveys useful information. We examine whether the relative prominence of individual risk factors within Item 1A reflects firms' exposure to the underlying risks and predicts future adverse outcomes. Focusing on credit and goodwill risk disclosures, we find that risk factor prominence is associated with proxies for the underlying risks and predicts credit rating downgrades, bankruptcy filings, and goodwill impairments. We further find that prominence is more informative during periods of high information uncertainty, when the benefits of risk disclosure are predicted to be greater. Overall, our findings suggest that risk factor prominence offers a valuable signal of firm risk that complements the textual disclosures in Item 1A. Accordingly, investors, analysts, lenders, auditors, boards, and regulators should consider both the level of, and changes in, risk factor prominence when evaluating firm risk.
- 16 March 2026
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 23 (1) , 107-126
Addressing the Last Mile Problem in Open Government Data: Using AIS Technologies to Enhance Governmental Financial Reporting
Although the OGD initiative has gained global momentum over the past two decades, the lack of a machine-readable format for much financial OGD in the U.S. hinders stakeholders' ability to use this information. This study draws on the “last mile problem”—a term that originally symbolizes inefficiencies in the final stage of delivering goods or services to end-users—to describe the difficulties that stakeholders face when analyzing PDF-type financial OGD. Following a design science methodology, this study proposes a report analysis framework to address this problem, develop processes, and evaluate its performance through a GASB standard-setting process (PIR). Results indicate that this framework achieves a 95.8 percent accuracy rate for data extraction from governmental reports and is four times faster than GASB's existing approach. This research contributes to the government accounting literature by applying accounting information system technologies to enhance the usability of OGD for various accounting users. JEL Classifications: M41; M42; M48; H83.
- 16 March 2026
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 23 (1) , 107-126
Addressing the Last Mile Problem in Open Government Data: Using AIS Technologies to Enhance Governmental Financial Reporting
Although the OGD initiative has gained global momentum over the past two decades, the lack of a machine-readable format for much financial OGD in the U.S. hinders stakeholders' ability to use this information. This study draws on the “last mile problem”—a term that originally symbolizes inefficiencies in the final stage of delivering goods or services to end-users—to describe the difficulties that stakeholders face when analyzing PDF-type financial OGD. Following a design science methodology, this study proposes a report analysis framework to address this problem, develop processes, and evaluate its performance through a GASB standard-setting process (PIR). Results indicate that this framework achieves a 95.8 percent accuracy rate for data extraction from governmental reports and is four times faster than GASB's existing approach. This research contributes to the government accounting literature by applying accounting information system technologies to enhance the usability of OGD for various accounting users. JEL Classifications: M41; M42; M48; H83.
- 8 September 2025
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 22 (2) , 125-147
Exploratory Experiments in Generating Bag of Word Dictionaries Using Word2Vec and ChatGPT
This paper investigates the use of machine learning as a means of generating “bag of words” using text corpora from accounting applications. We use Word2Vec to generate words/dictionaries, that are “similar” to a seed word that captures a concept. As part of our analysis, we perform several experiments using text from Form 10Ks and earnings calls. We investigate several activities including choice of the seed word(s), choosing word sources (corpora), analysis of resulting word lists, and other concerns. We also examine the notion of “human-in-the-loop” and the roles that a person would need to perform while generating a dictionary. Further, we investigate the impact of using accounting and financial corpuses on the different semantic and syntactic relationships, in contrast to Wikipedia. We then extend the analysis to compare those findings to ChatGPT another source of words and investigate some of the advantages and disadvantages of that approach. Data Availability: Data are available from the authors. JEL Classifications: M4.
- 8 September 2025
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 22 (2) , 125-147
Exploratory Experiments in Generating Bag of Word Dictionaries Using Word2Vec and ChatGPT
This paper investigates the use of machine learning as a means of generating “bag of words” using text corpora from accounting applications. We use Word2Vec to generate words/dictionaries, that are “similar” to a seed word that captures a concept. As part of our analysis, we perform several experiments using text from Form 10Ks and earnings calls. We investigate several activities including choice of the seed word(s), choosing word sources (corpora), analysis of resulting word lists, and other concerns. We also examine the notion of “human-in-the-loop” and the roles that a person would need to perform while generating a dictionary. Further, we investigate the impact of using accounting and financial corpuses on the different semantic and syntactic relationships, in contrast to Wikipedia. We then extend the analysis to compare those findings to ChatGPT another source of words and investigate some of the advantages and disadvantages of that approach. Data Availability: Data are available from the authors. JEL Classifications: M4.
- Top Cited Papers
- 1 October 2023
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 20 (2) , 1-10
Large Language Models: An Emerging Technology in Accounting
This commentary discusses how large language models like ChatGPT hold transformative potential in accounting, including education, research, and professional auditing. In the educational sphere, the advent of ubiquitous artificial intelligence (AI) tutors could potentially solve Bloom’s Two Sigma Problem, heralding a new era of personalized learning. Accounting research stands to benefit immensely, particularly in tasks that rely heavily on natural language processing. In the professional auditing domain, the capabilities of ChatGPT to create broad outlines of risks inherent in certain accounts and assertions can enable engagement teams to create more risk-responsive audit plans. However, although the advantages are remarkable, they are accompanied by potential pitfalls that necessitate cautious navigation. Even with these challenges, AI’s impending transformation in personal and professional lives cannot be overlooked, as accounting stands on the brink of significant change. JEL Classifications: M40; M42; O33.
- 1 October 2023
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 20 (2) , 1-10
Top Cited PapersLarge Language Models: An Emerging Technology in Accounting
This commentary discusses how large language models like ChatGPT hold transformative potential in accounting, including education, research, and professional auditing. In the educational sphere, the advent of ubiquitous artificial intelligence (AI) tutors could potentially solve Bloom’s Two Sigma Problem, heralding a new era of personalized learning. Accounting research stands to benefit immensely, particularly in tasks that rely heavily on natural language processing. In the professional auditing domain, the capabilities of ChatGPT to create broad outlines of risks inherent in certain accounts and assertions can enable engagement teams to create more risk-responsive audit plans. However, although the advantages are remarkable, they are accompanied by potential pitfalls that necessitate cautious navigation. Even with these challenges, AI’s impending transformation in personal and professional lives cannot be overlooked, as accounting stands on the brink of significant change. JEL Classifications: M40; M42; O33.
- 1 July 2019
- Published by Wiley in Expert Systems
- Vol. 36 (4)
Gathering and evaluating innovation ideas using crowdsourcing: Impact of the idea title and the description on the number of votes in each phase of a two‐phase crowdsourcing project
Organizations are using crowdsourcing to capture innovation knowledge from the crowd in the form of ideas and then using the crowd to evaluate those ideas using votes. In this paper, we investigate a crowdsourcing setting in which Canada solicited information from its citizens to develop a digital transformation strategy. Canada used a two‐phase approach. Phase 1 was used to determine which ideas had the largest number of crowd votes, whereas in Phase 2, the crowd voted on the 30 leading vote‐getting ideas to determine the three winning ideas.This research investigates the ability to use information from ideas to estimate the number of votes that the ideas generate. This approach could be used to estimate the number of ideas, before making information available to the crowd. The unstructured text information in the idea is structured by using target concept dictionaries, which are used to estimate the extent to which the dictionary words appear in the ideas (e.g., “globalism”) and are related to the number of votes. Using this approach, roughly 1% of the total words are used to explain roughly 60% of the variance in the votes. Further, we also find that the variables associated with Phase 1 votes are not the same variables associated with Phase 2 votes; that is, the decision‐making variables changed. Finally, we find that votes are statistically significantly related to the content in the idea titles and the idea statements.
- 1 July 2019
- Published by Wiley in Expert Systems
- Vol. 36 (4)
Gathering and evaluating innovation ideas using crowdsourcing: Impact of the idea title and the description on the number of votes in each phase of a two‐phase crowdsourcing project
Organizations are using crowdsourcing to capture innovation knowledge from the crowd in the form of ideas and then using the crowd to evaluate those ideas using votes. In this paper, we investigate a crowdsourcing setting in which Canada solicited information from its citizens to develop a digital transformation strategy. Canada used a two‐phase approach. Phase 1 was used to determine which ideas had the largest number of crowd votes, whereas in Phase 2, the crowd voted on the 30 leading vote‐getting ideas to determine the three winning ideas.This research investigates the ability to use information from ideas to estimate the number of votes that the ideas generate. This approach could be used to estimate the number of ideas, before making information available to the crowd. The unstructured text information in the idea is structured by using target concept dictionaries, which are used to estimate the extent to which the dictionary words appear in the ideas (e.g., “globalism”) and are related to the number of votes. Using this approach, roughly 1% of the total words are used to explain roughly 60% of the variance in the votes. Further, we also find that the variables associated with Phase 1 votes are not the same variables associated with Phase 2 votes; that is, the decision‐making variables changed. Finally, we find that votes are statistically significantly related to the content in the idea titles and the idea statements.
- 1 June 2019
- Published by American Accounting Association in Accounting Horizons
- Vol. 33 (3) , 111-126
Contract Analytics in Auditing
SYNOPSIS: This paper proposes the Contract Analytics Framework (CAF), which provides guidance for incorporating textual analysis into audits of large numbers of contracts. Contracts are used extensively in auditing for risk assessment, analytical procedures, substantive testing, and audit review. While material contracts are carefully examined to assess financial implications, contracts perceived to be low risk are typically audited using standard sampling techniques. This proposed CAF facilitates effective and efficient audit analyses on full populations of contracts. We identify and describe six functional areas in the CAF: (1) Document Management, (2) Content Identification, (3) Cutoff Testing, (4) Record Confirmation, (5) Term Verification, and (6) Additional Audit Tasks. Based on auditing standards, we discuss the application of the CAF in the audit stages of risk assessment, substantive tests, and review. The framework is then implemented on a group of reinsurance contracts to demonstrate the feasibility of auditing full populations of contracts. JEL Classifications: M42; C88; O39.
- 1 June 2019
- Published by American Accounting Association in Accounting Horizons
- Vol. 33 (3) , 111-126
Contract Analytics in Auditing
SYNOPSIS: This paper proposes the Contract Analytics Framework (CAF), which provides guidance for incorporating textual analysis into audits of large numbers of contracts. Contracts are used extensively in auditing for risk assessment, analytical procedures, substantive testing, and audit review. While material contracts are carefully examined to assess financial implications, contracts perceived to be low risk are typically audited using standard sampling techniques. This proposed CAF facilitates effective and efficient audit analyses on full populations of contracts. We identify and describe six functional areas in the CAF: (1) Document Management, (2) Content Identification, (3) Cutoff Testing, (4) Record Confirmation, (5) Term Verification, and (6) Additional Audit Tasks. Based on auditing standards, we discuss the application of the CAF in the audit stages of risk assessment, substantive tests, and review. The framework is then implemented on a group of reinsurance contracts to demonstrate the feasibility of auditing full populations of contracts. JEL Classifications: M42; C88; O39.
- 1 September 2018
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 15 (2) , 67-75
A Framework for Legacy Source Code Audit Analytics
An internal audit engagement at a large bank revealed that legacy software code can represent significant risk to an organization. In this engagement, several instances of malicious COBOL source code were discovered through manual inspection and identification. The internal auditors were concerned that more instances of faulty code could be in the production environment, and that manually inspecting over 100 million lines of source code would be infeasible. Under their auspices, I developed the Retrospective Static Source Code Analysis Framework for automatically identifying source code files likely to contain source code faults. We implemented this audit analytic framework in a tool called COBOL Analyzer. Using the tool, the auditors discovered thousands of potential source code files with faults and, using a ranking method, audited the most suspicious ones. The tool was deemed a success and it is undergoing further development by the internal audit team.
- 1 September 2018
- Published by American Accounting Association in Journal of Emerging Technologies in Accounting
- Vol. 15 (2) , 67-75
A Framework for Legacy Source Code Audit Analytics
An internal audit engagement at a large bank revealed that legacy software code can represent significant risk to an organization. In this engagement, several instances of malicious COBOL source code were discovered through manual inspection and identification. The internal auditors were concerned that more instances of faulty code could be in the production environment, and that manually inspecting over 100 million lines of source code would be infeasible. Under their auspices, I developed the Retrospective Static Source Code Analysis Framework for automatically identifying source code files likely to contain source code faults. We implemented this audit analytic framework in a tool called COBOL Analyzer. Using the tool, the auditors discovered thousands of potential source code files with faults and, using a ranking method, audited the most suspicious ones. The tool was deemed a success and it is undergoing further development by the internal audit team.