Introduction

Auditing government construction projects is confronted with inherently complex and multidimensional challenges, ranging from issues of data complexity to limited transparency. These conditions demand auditors to employ systematic, comprehensive, and technology-enabled approaches to ensure accountability, detect irregularities, and enhance the overall effectiveness of the audit process. In government construction projects, understanding the characteristics and challenges of construction data becomes crucial, fundamentally determine the effectiveness of technology-driven audit approaches. Construction project data are typically characterized by high volume, heterogeneous formats, and dynamic structure (Aouad et al., 1999). In the construction sector, data arise from heterogeneous sources such as design systems, project scheduling tools, Enterprise Resource Planning (ERP) applications, and financial records, leading to massive data volumes ranging from terabytes to petabytes (Bilal et al., 2016). The acquisition of construction data remains particularly challenging due to persistent issues such as manual data entry, dynamic and static occlusions, and the absence of uniform data standards (Han & Golparvar-Fard, 2015; Seo et al., 2015). In the construction industry, the operation of data platforms still relies on information exchange among different domains—such as structural, architectural, electrical, and mechanical—which often encounter interconnection challenges (Yousif et al., 2021).

Recent breakthroughs in artificial intelligence (AI) have begun to transform industrial practices worldwide. Nevertheless, the construction industry has been comparatively slow in adopting these technologies (Ghimire et al., 2024). Ghimire et al., (2024) highlights several challenges in the application of machine learning within the construction sector. A challenge occurs when owner organizations attempt to implement machine learning models for predicting construction project completion time based on available data, such as project value, delivery method, complexity, and material quantities. The key issue is the trade-off between predictive accuracy and generalizability - models with higher precision yield more accurate predictions but may lack broad applicability, while less precise models are more generalizable but sacrifice accuracy.

The challenges discussed above are predominantly related to the technical aspects of construction. Equally important, however, are the non-technical aspects—such as corruption and fraud in government construction projects—which also warrant careful consideration. Corruption has been identified as a major problem in construction projects (Zhai et al., 2021). Zarghami (2025) study indicates that the four elements of the fraud diamond theory can trigger corrupt behavior in construction projects. This finding highlights that fraudulent patterns present significant challenges for detection through conventional data analysis. Consequently, there is an increasing need for systems capable of identifying anomalies and detecting fraud patterns in government construction projects. In this context, leveraging advanced technological solutions, particularly BDA and GenAI, emerges as a strategic approach warranting serious consideration.

The purpose of this article is to examine the potential application of BDA and GenAI as transformative solutions for enhancing the effectiveness and efficiency of construction project audits, particularly in detecting anomalies and patterns of fraud. To the best of the author’s knowledge, no prior study has specifically examined the linkage between the use of BDA and GenAI and the detection of anomalies and fraud in government construction projects. This study contributes to advancing the discourse by offering insights and recommendations for audit institutions to adopt an audit approach using BDA and GenAI that is adaptive to technological advancements and the evolving governance challenges of modern public construction projects.

The Essence of BDA and GenAI in Government Construction Audits for Detecting Anomalies and Fraud Patterns

GenAI, capable of producing new and realistic data or content based on given inputs or existing knowledge, offers innovative solutions to challenges across various processes in the construction industry, including design, planning, procurement, inspection, and maintenance (Taiwo et al., 2025). Recent studies Alwashah et al., (2025) indicate that the construction sector is beginning to explore GenAI to address complex data management challenges and inefficiencies inherent in traditional workflows, particularly in design, planning, and project management. These studies conclude that GenAI technologies play a pivotal role in construction.

Considering this potential, the application of GenAI is not only relevant for enhancing efficiency and accuracy in technical construction processes but also offers strategic opportunities not only anomaly detection but also fraud prevention and identification in government construction projects. Fraud detection and prevention are critical in construction projects, with common U.S. industry frauds including billing schemes, corruption, cost substitution, and non-cash misappropriation (Ward, 2024). Traditional detection methods often fall short against increasingly sophisticated fraud. GenAI can provide visualizations and narratives from fraud data (Ward, 2024).

The study by Rosnidah et al., (2022) demonstrates that BDA substantially enhances auditing processes, as leveraging big data as supplementary audit evidence facilitates the identification of anomalies and the prediction of fraudulent activities. Data-driven approaches to detecting fraud in public procurement have received considerable scholarly attention Santos et al., (2025), encompassing techniques such as Machine Learning (ML) for cartel identification (Wallimann & Sticher, 2023) and advanced statistical methods for the detection of anomalous bidding behaviors (Maia et al., 2020). Santos et al., (2025) research highlights that the study of red flags for collusion detection is comparatively underexplored relative to other forms of fraud, and emphasizes the need for the development of robust, data-driven anti-fraud systems.

Best Practices in Utilizing BDA and GenAI in Construction and Audit

To the best of the author’s knowledge, there has been no comprehensive and specific example demonstrating the successful use of BDA and GenAIfor detecting anomalies and fraud in government construction projects by audit institutions. Nonetheless, several cases illustrate the use of BDA and GenAI in public construction management or in anomaly and fraud detection in general, although not specifically within government construction projects. In Indonesia, The Toll Road Regulatory Agency (BPJT) has partnered with researchers from Universitas Gadjah Mada (UGM) to apply AI in monitoring toll road conditions across Indonesia. This technological innovation enhances the maintenance of toll road infrastructure assets, as BPJT oversees approximately 2,300 kilometers of toll roads operated by 40 state-owned and private companies. The implementation of AI is anticipated to deliver real-time information on potential cracks or potholes, enabling timely and proactive maintenance actions. According to Munawar et al., (2022), the adoption of tools such as Computer-Aided Design (CAD) and Building Information Modeling (BIM) offers significant opportunities for researchers in the construction industry to enhance how infrastructure can be developed, monitored, and improved in the future. Big Data Engineering (BDE) demonstrates remarkable applications within construction (Munawar et al., 2022), having been employed alongside BIM to advance project management (Huang, 2021). BDE has also been utilized to optimize building design and performance monitoring (Loyola, 2018), project management, safety, energy management, decision-making frameworks for design, resource management (Ismail et al., 2018), quality control, waste management, and other areas (Wang et al., 2018). Furthermore, Zabala-Vargas et al., (2023) notes that emerging technologies—including big data, data science, and AI—have become viable alternatives throughout the project lifecycle. The most notable contributions of AI relate to project development forecasting, identification of critical factors, detailed risk assessment, planning optimization, task automation, and efficiency enhancement, thereby facilitating more informed management decision-making. Meanwhile, Rodríguez et al., (2022) evaluated the accuracy of eleven ML algorithms for detecting collusion using datasets from Brazil, Italy, Japan, Switzerland, and the United States. The findings suggest that ML holds significant promise for identifying collusive behavior. The algorithms are highly flexible, achieving reasonable detection rates even with limited information.

According to Álvarez-Foronda et al., (2023), such tools not only advance fraud detection but also enable the evolution of audit processes, enhancing both their effectiveness and efficiency. These technologies facilitate near real-time automation of testing and risk identification, prompting a shift toward a continuous audit model as opposed to traditional static auditing approaches. Data analytics as an audit tool allows for reduced labor time and optimized allocation of effort across audit phases (efficiency), such as extending audit coverage and extrapolating findings to entire populations rather than relying on traditional sampling methods (effectiveness). The adoption of IT-enabled auditing techniques has been shown to improve overall audit efficiency and effectiveness. Furthermore, Ismail et al., (2024) emphasizes that data analytics in auditing has substantial potential to enhance audit quality, minimize errors, increase process transparency, and strengthen stakeholder credibility. The Audit Board of the Republic of Indonesia (BPK) has begun developing and implementing BDA through its BIDICS platform as part of the digitalization of the audit process. This platform is designed to collect, integrate, and analyze large and diverse datasets relevant to public sector audits, including both financial and non-financial data, while also providing analytical support for selected specialized audits. The adoption of BDA within BPK has significantly expanded the scope of audit testing and enhanced the depth of analytical insights.

Challenges in Utilizing BDA and GenAI in Government Construction Audits for Detecting Anomalies and Fraud Patterns

The application of BDA and GenAI in government construction audits presents several challenges. According to Ismail et al., (2024), the challenges associated with the use of data analytics in auditing include large data volumes, a lack of standardized frameworks, and negative perceptions. Meanwhile, research by Kokina et al., (2025) indicates that “simple AI” technologies, such as key data extraction from documents and optical character recognition, are widely employed in auditing, whereas more advanced “complex AI” tools are still under development. The study further reports that Robotic Process Automation (RPA) is commonly used to automate repetitive administrative tasks, but its application in audit-specific functions remains limited. Additionally, Kokina et al., (2025) identifies key barriers to AI adoption in auditing, including issues of transparency and explainability, algorithmic bias, data privacy, system resilience and reliability, auditors’ concerns over excessive reliance on AI, and the need for comprehensive AI guidelines.

This aligns with Ward (2024), which notes that the use of GenAI for fraud detection and prevention in construction faces several challenges, including data quality and availability. GenAI relies on large and diverse datasets to train and evaluate models, yet construction data are often scarce, incomplete, and/or inconsistent due to the heterogeneous nature of the industry. Moreover, construction data are frequently sensitive and confidential, and accessing or sharing such data may raise privacy and security concerns. Ward (2024) further highlights challenges related to model validity and reliability. GenAI models are often complex, making their outputs difficult to interpret and verify, which raises questions regarding the accuracy, reliability, and suitability of the models for fraud detection and prevention purposes. Finally, Ward (2024) emphasizes regulatory and ethical considerations. The deployment of GenAI can have legal and ethical implications, particularly when used to inform decisions affecting stakeholders’ rights and interests. Consequently, the application of GenAI for fraud detection and prevention in construction requires compliance with relevant laws and regulations.

Conclusion

The integration of BDA and GenAI into government construction auditing represents a paradigm shift from traditional audit practices toward an intelligent, data-driven, and adaptive audit model. This transformation enables auditors to extend their analytical scope from limited data samples to large-scale datasets, thereby enhancing the validity while minimizing the risk of interpretive bias. The adoption of BDA and GenAI not only improves audit efficiency and effectiveness but also reinforces public accountability through greater transparency, reliability of evidence, and objectivity in decision-making. These technologies facilitate the early and comprehensive detection of anomalies and fraudulent patterns within complex construction project data environments.

Despite their promising potential, the successful implementation of BDA and GenAI in the audit of public construction projects hinges on overcoming several critical challenges, data quality issues, limited technical capacity among auditors, insufficient data infrastructure, and the absence of regulatory frameworks. From an academic perspective, this article contributes to the conceptual advancement of BDA and GenAI utilization in government construction auditing. It also lays the groundwork for future research on broader examples of successful AI-based anomaly detection and the strengthening of public-sector internal control systems oriented toward fraud prevention, particularly in the management and oversight of government construction projects.