Artificial Intelligence and Information Technology in Audit Assurance: Global Trends and Nepal's Evolving Landscape

Artificial Intelligence and Information Technology in Audit Assurance: Global Trends and Nepal's Evolving Landscape

By: Aayush Chaudhary

Audit & Assurance | Artificial Intelligence | Information Technology | Nepal

Executive Summary

The global audit profession is undergoing a structural transformation driven by artificial intelligence (AI) and advanced information technology.

Audit firms worldwide are moving beyond sampling-based procedures toward full-population analysis, real-time risk detection, and agentic AI systems that plan, execute, and document audit work under human oversight.

These developments are directly relevant to the core assurance objective: providing credible, reliable information to third parties — investors, lenders, regulators, and the public — who rely on audited financial statements for decision-making.

In Nepal, the adoption trajectory is markedly different. While the Institute of Chartered Accountants of Nepal (ICAN) has introduced certification courses on AI and strategic plans acknowledge digital transformation, the practical integration of AI into audit workflows remains at an early stage.

This article examines global developments and recent adaptations, assesses their alignment with Nepal Standards on Auditing (NSA), and evaluates Nepal's readiness in comparative perspective.

1. The Global Shift: From Sampling to Full-Population Assurance

1.1 The Scale of Transformation

One of the most consequential changes in audit methodology is the shift from sampling to full-population analysis.

AI models can now scan every transaction a company makes annually rather than relying only on a limited sample. This significantly expands the evidence base supporting an audit opinion.

The AI-enabled audit lifecycle can be viewed through four broad stages:

1. Risk assessment

2. Full-population analytics

3. Anomaly and evidence review

4. Human judgment and conclusion

AI scales the analysis, but the auditor retains professional scepticism and responsibility.

According to the research presented in this article, IFIAR reported that in 2025, Global Public Policy Committee (GPPC) networks collectively reported an approximate 30% year-over-year increase in AI-enabled tools.

Gartner's survey of chief audit executives found that 83% of audit functions were piloting or using AI, with another 12% planning to follow within the year. A survey by SAI Egypt found that 92% of respondents agreed that AI enhances audit results, while 87% supported its use in risk assessments.

1.2 Agentic AI: The 2025–2026 Inflection Point

Agentic AI refers to systems capable of planning, executing, and summarising procedures autonomously.

According to the article's research, agentic AI has moved from experimentation toward engagement use.

KPMG began integrating agents into its Clara audit platform in April 2025, initially focusing on areas such as expense vouching and searches for unrecorded liabilities.

EY embedded a Microsoft-built multiagent framework in its Canvas platform in April 2026, with the objective of achieving end-to-end coverage by 2028 across 130,000 assurance professionals in more than 150 countries.

Deloitte's Zora and PwC's Agent OS follow similar design principles. PwC Australia is also rolling out an AI-native audit platform across 1,600 auditors, with implementation targeted by 2028.

1.3 Human-in-the-Loop as a Design Principle

A critical feature of global AI adoption is the deliberate use of human oversight.

The human-in-the-loop assurance model can be understood as:

AI → identifies data patterns and potential risks

Auditor → reviews and validates evidence

Auditor → applies professional judgment and professional scepticism

Auditor → reaches the final conclusion

Global audit-firm approaches emphasise that AI can identify potential risks and suggest procedures, but the auditor remains responsible for evaluating whether those procedures are appropriate and for reaching the final conclusion.

This preserves professional scepticism, judgment, and accountability while allowing technology to expand the scale and speed of audit analysis.

2. Alignment with Nepal Standards on Auditing (NSA)

2.1 The NSA Framework

The Nepal Standards on Auditing (NSA) are based on the International Standards on Auditing (ISA) and are required to be applied in statutory audits.

The article notes that the Auditing Standards Board of Nepal revised NSA in 2024, with voluntary application beginning in July 2024 and mandatory application from 16 July 2025.

The standards are complemented by:

• Nepal Standards on Assurance Engagements (NSAE)

• Nepal Standards on Quality Control (NSQC)

• The Information System Audit (ISA) Manual published by ICAN

2.2 How AI-Enhanced Procedures Align with NSA Requirements

AI-enhanced audit procedures can support several key audit requirements.

NSA 315, relating to understanding the entity and its environment, requires auditors to identify and assess risks of material misstatement.

AI-driven predictive analytics and real-time data monitoring can support this process by identifying unusual or unauthorised transactions that traditional sampling may not detect.

NSA 240, relating to fraud, requires procedures to detect material misstatement due to fraud. Machine-learning models that identify anomalous patterns can directly support this objective.

NSA 500, relating to audit evidence, requires sufficient appropriate evidence. Full-population testing can provide a more comprehensive evidentiary basis than traditional sampling.

The ISA Manual published by ICAN explicitly addresses Computer Assisted Audit Techniques/Tools (CAATs) and data analytics in auditing, providing a framework for integrating technology into NSA-compliant audits.

The Quality Assurance Board's Audit Practice Manual (APM) is designed as an NSA- and Code-of-Ethics-compliant audit system and is recommended for use by practising audit firms in Nepal.

2.3 Gaps and Tensions

Despite these areas of alignment, the current NSA framework does not explicitly address several AI-specific risks.

These include:

• Algorithmic opacity

• Over-reliance on AI outputs without sufficient challenge

• Reliability of third-party data sources embedded in AI models

• Governance of AI-generated audit work

IFIAR has emphasised the need for firms to evolve and re-engineer processes using International Standard on Quality Management (ISQM) 1 to ensure effective monitoring and oversight of technology use in audits.

Nepal's NSQC framework could benefit from more explicit guidance on AI governance, explainability, and the evidentiary status of AI-generated workpapers.

3. Enhancing Third-Party Decision-Making

3.1 The Assurance Value Proposition

The fundamental purpose of audit assurance is to reduce information asymmetry between company management and third parties such as investors, creditors, regulators, and the public.

AI and information technology can enhance this assurance value in three important ways.

Expanded Evidence Coverage

AI-enabled platforms can analyse large or complete populations of transactions, expanding the scope of scrutiny applied during an audit.

For example, an AI-native audit platform can provide extensive analysis of judgments and transactions that are critical to the audit process.

For a lender assessing a borrower's financial statements, broader analysis can reduce detection risk.

Real-Time Risk Identification

AI systems can identify anomalies and high-risk areas in real time, enabling auditors to address potential problems before they develop into material misstatements.

AI agents can assist in discovering anomalies, tracking trends, and identifying risks within the data being audited.

Enhanced Comparability and Consistency

AI agents can apply consistent analytical logic to comparable data.

This can support a more standardised approach to risk assessment and audit responses across clients and sectors.

Greater consistency can enhance the reliability of audit opinions and benefit third parties who rely on financial information for comparison and decision-making.

3.2 The Reliability Caveat

AI adoption also introduces important risks.

IFIAR has cautioned against over-reliance on AI outputs without sufficient challenge of the work performed.

The central concern is not only whether AI replaces professional judgment, but whether practitioners may delegate an inappropriate share of their work to AI while retaining formal responsibility for the result.

For third parties, this creates a central tension:

AI can expand the scope of assurance, but effective governance is necessary to ensure that the quality of assurance is not compromised.

Therefore, technological capability must be accompanied by professional judgment, review, validation, and appropriate governance.

4. Nepal's Position: Aspiration Meets Infrastructure Reality

4.1 Institutional Momentum

Nepal's professional institutions have begun responding to the global shift toward AI and digital auditing.

ICAN's Strategic Plan 2024/25–2028/29 identifies technology and digitalization as important areas and acknowledges the gradual adoption of digital tools in Nepal.

The strategy positions ICAN to contribute to digital transformation in financial reporting and auditing practices.

ICAN has also launched a Certification Course on Artificial Intelligence in technical collaboration with the Institute of Chartered Accountants of India (ICAI), with courses held in October 2025 and August 2026.

A certificate program on data analytics and AI has also been launched, with orientation provided to members on responsible AI use.

The Nepal Auditors Association (ADAN) has also placed AI among its areas of professional discussion, focusing on professional capacity, AI use, tax reform, and skilled human-resource development.

These developments indicate institutional movement toward preparing the profession for technological change.

4.2 The Office of the Auditor General and Public Sector Technology

The Office of the Auditor General of Nepal (OAGN) has made more concrete progress in applying technology to auditing.

The Nepal Audit Management System (NAMS), developed with Cowater, modernised OAGN's financial audit methodologies, aligned them with ISSAIs, and introduced computer-assisted auditing techniques.

OAGN uses optical character recognition and reconciliation algorithms for revenue collection data to support audit conclusions.

Nepal Rastra Bank has also issued AI guidelines requiring banks and financial institutions to report AI system usage, risk management, and cybersecurity frameworks annually.

These represent meaningful steps, although the article notes that such developments remain concentrated in the public sector and regulated financial institutions.

4.3 The Private Audit Firm Reality

The situation in Nepal's private audit-firm sector is different.

The article notes that many Nepali SMEs continue to rely on manual bookkeeping, Tally, or Excel, while AI usage remains very low compared with developed countries.

The Nepal AI in Accounting Market report identifies several barriers, including:

• Limited awareness

• Data-security concerns

• Lack of skilled professionals

AI-native firms such as Sterling Wells Nepal are identified in the article as exceptions.

The article also reports that Nepal ranks 150th out of 193 nations on the global AI Readiness Index, highlighting infrastructure limitations and shortages in localised data-verification capabilities.

This creates a significant contrast between global audit firms deploying agentic AI and many Nepali audit firms that are still working toward digitising basic working papers.

4.4 Comparative Analysis

The comparison illustrates the current difference between the maturity of AI-enabled audit practices globally and the developing state of adoption within Nepal's private audit sector.

5. Preferences and Recent Adaptations: What Nepal Can Learn

5.1 Preference for Human-Led, AI-Augmented Models

The global direction described in the article favours a human-led, AI-augmented approach.

AI should be used to deepen analysis, identify patterns, and support critical thinking rather than replace professional judgment.

Nepal's adoption strategy can therefore prioritise AI as an augmentation tool for professional judgment.

ICAN's emphasis on foundational understanding and practical AI skills is aligned with this approach.

5.2 Investment in Explainability and Governance

Global firms have developed greater transparency and audit logging within their AI platforms to support algorithmic explainability and regulatory compliance.

Nepal's regulatory framework, including NSA 2024 and NSQC, could be supplemented with explicit guidance on AI governance.

Such guidance could address:

• Validation of AI-generated evidence

• Mitigation of over-reliance

• Assessment of third-party data sources

• Documentation of AI use

• Explainability and audit trails

5.3 Scaling Through Shared Platforms

Building proprietary AI platforms can be prohibitively expensive for many Nepali audit firms.

The OAGN's NAMS model, which uses a shared government-procured audit management system, provides a potential template.

ICAN could facilitate the development of a shared cloud-based audit platform for small and medium practitioners.

Such a platform could incorporate:

• NSA-compliant workflows

• Computer Assisted Audit Techniques (CAATs)

• Data analytics

• Secure cloud workflows

• AI governance mechanisms

The existing Audit Practice Manual could provide a foundation for such development.

5.4 Data Quality as the Binding Constraint

One of the most important challenges in adopting AI for auditing is data quality.

Gartner's survey cited in the article indicates that AI outcomes can be limited by poor data quality, lack of technical development skills, and restricted access to technology.

This issue is particularly relevant in Nepal, where manual bookkeeping remains common among SMEs.

AI-enhanced auditing cannot compensate for unstructured or unreliable client data.

Therefore, investment in digital bookkeeping infrastructure at the client level is a prerequisite for meaningful AI adoption on the audit side.

6. A Practical AI-Assurance Roadmap for Nepal

The article proposes a sequenced approach to developing AI-enabled assurance in Nepal:

Step 1: Digital Client Data

Build stronger foundations through:

• Digital bookkeeping

• ERP systems

• Structured financial records

Step 2: Shared Audit Tools

Develop accessible technology incorporating:

• CAATs

• Data analytics

• Secure cloud workflows

Step 3: AI Governance

Establish mechanisms for:

• Validation

• Explainability

• Logging

• Professional oversight

Step 4: Professional Skills

Invest in:

• AI literacy

• Data analytics

• NSA-focused training

• Continuing professional development

This sequencing recognises that effective AI auditing requires more than simply acquiring AI software. It requires quality data, appropriate infrastructure, governance, and skilled professionals.

Conclusion

The global audit profession is in the midst of a technology-driven transformation that is materially enhancing the assurance available to third parties.

Full-population testing, real-time risk detection, and agentic AI under human oversight are expanding the scope and timeliness of audit evidence in ways that can benefit investors, lenders, and regulators.

These developments are broadly consistent with the objectives of the Nepal Standards on Auditing, which are ISA-based and principle-oriented. However, the current NSA framework does not contain explicit AI-specific guidance.

Nepal's position can be characterised by institutional awareness and early-stage capability-building.

ICAN's certification courses, OAGN's NAMS implementation, and Nepal Rastra Bank's AI guidelines represent important steps. At the same time, the private audit-firm sector remains substantially manual and sampling-based.

Closing this gap requires a sequenced approach:

1. Strengthening data quality and digital bookkeeping at the client level

2. Building shared audit-technology infrastructure accessible to small and medium firms

3. Developing NSA-specific AI governance guidance

4. Continuing investment in professional education

The assurance value proposition — providing credible information for third-party decision-making — depends on the profession's ability to evolve its methods alongside the increasing complexity of the information environment it audits.

AI is not a threat to the assurance function; under proper governance, it can become a powerful enabler of it.