South Africa and Nigeria Drive AI-Enabled Transfer Pricing Audits

AI is changing the practice of tax law. This series examines the ethical, legal, and practical implications of AI across key areas of tax practice.

Across Africa, South Africa and Nigeria stand out as two of the most sophisticated jurisdictions when considering the level of transfer pricing audit maturity, not only because of the frequency and scope of audits, but also due to the complexity of multinational groups operating within their borders. Consequently, both countries have developed advanced transfer pricing compliance methodologies characterized by in-depth reviews, extensive documentation requirements, and rigorous testing of the economic substance underlying cross-border intercompany transactions.

Transfer pricing audits continue to evolve as tax authorities worldwide pursue transparency, accuracy, and fairness in the allocation of profits across jurisdictions. This evolution is shaped by a combination of structural and geopolitical forces. Global value chains are expanding while becoming increasingly fragmented, profit allocation rules are being re-examined, reporting obligations are intensifying, and domestic tax laws are undergoing continuous reforms. Together, these developments are reshaping the transfer pricing landscape and prompting tax authorities to adopt more sophisticated technologies in assessing cross-border related party transactions.

While tax authorities have been using risk assessment methodologies and tools in Africa and the rest of the world for a while, for example, making use of information set out in country-by-country reports, or the Transfer Pricing Risk Assessment Tool, developed by the African Tax Administration Forum and released in 2024, these methodologies and tools relied on a somewhat manual intervention process. With recent technologies, particularly artificial intelligence-based solutions, such traditional methodologies may become more advanced.

Beyond structural and geopolitical drivers, technology has become an equally powerful force shaping transfer pricing audit practices. Generative artificial intelligence, or GenAI, and related technologies are transforming how transfer pricing audits are conducted. AI-powered risk engines can analyze vast datasets at high speed, while natural language processing tools enable tax authorities to review large volumes of transfer pricing documentation, benchmarking studies, and financial disclosures far more efficiently than before.

This transformation raises a critical question: as AI becomes increasingly embedded in transfer pricing audit processes, could it ultimately tilt the playing field against taxpayers?

This article examines how the risks associated with AI‑enabled audits, including the risk of data misinterpretation, can be mitigated. It also provides guidance for tax policymakers and administrators on how to design robust governance frameworks before AI‑enabled transfer pricing audits become fully operational. Finally, it outlines practical steps multinational enterprises, should take to prepare strategically and operationally.

South Africa

The South African Revenue Service has invested, and continues to invest, heavily in technology and digital transformation. It therefore would not be surprising if the administration achieved noteworthy progress in integrating advanced technologies, particularly GenAI, into the transfer pricing audit process. While the revenue service has, to date, not openly confirmed the use of GenAI in the tax audit process, the organization uses AI chatbots to assist taxpayers, and it has been reported that AI tools are used for fraud detection. The developments are accelerating, and what we see today is likely only a glimpse of what is to come.

AI in taxpayer selection and risk assessment. With the advent of country-by-country reporting, large taxpayers were required to disclose certain information relating to their tax positions in a report that was (generally) provided to the tax authority in the country where the ultimate parent entity was tax resident. The reason for this disclosure was to enable a tax authority to perform a high-level risk assessment, and to share the information with other tax authorities in countries where the multinational group operated, provided there was a suitable exchange of information mechanism in place for the same purpose.

More recently, Public Country-by-Country Reporting requirements offer tax authorities in jurisdictions with exchange of information arrangement in place to nevertheless obtain relevant information that can be used for risk assessment. The use of AI has elevated this risk assessment to a new level.

Systems used for risk assessment are designed to analyze large, complex datasets with minimal human intervention and identify patterns that traditional review methods may miss. Early applications suggest that AI models can flag potentially high-risk multinationals by using natural language processing, or NLP, to evaluate the consistency of financial statements, annual income tax returns for companies and close corporations, submissions, transfer pricing documentation, and intercompany agreements.

AI in audit initiation and document review. Once a taxpayer is flagged as high-risk at the stage of the transfer pricing risk assessment, the process typically transitions to the audit initiation phase through issuing an audit letter and subsequently a request for relevant material. This phase involves detailed review of the taxpayer’s records, disclosures, and supporting documentation. In many cases, the revenue service must examine large volumes of agreements, invoices, memos, policies, and financial schedules, making it one of the most administrative phases of the audit.

AI and related technologies significantly reduce this burden through automated document extraction and classification. Using optical character recognition combined with NLP, AI systems can “read” contracts and highlight clauses related to pricing, service scope, intellectual property, and risk allocation, improving both speed and accuracy.

AI in the substantive review of intercompany transactions. The heart of a transfer pricing audit lies in determining whether intercompany transactions were conducted at arm’s length. This assessment often takes shape during the business review stage and the subsequent individual interviews, when revenue service officials engage directly with key personnel to understand the taxpayer’s business operations, the economic substance of cross-border intercompany transactions, and the actual functions performed, assets used, and risks assumed by the parties to the transactions. As the South African revenue service continues its digital transformation, AI can increasingly enhance this stage of the audit by enabling more structured functional analyses, supporting the conduct of interviews, strengthening company profiling, assisting with the selection of comparable transactions, and improving the quality of preliminary economic methodologies.

Emerging AI dilemma. Taxpayers have increasingly expressed dissatisfaction with the use of AI for risk profiling, interviews, information capture, and audit adjustments. The concern stems from the possibility that AI-generated inferences may fail to reflect the true economic reality of a taxpayer’s business structure, functions, and risks. Transfer pricing, by its nature, is a judgment intensive discipline. Determining an arm’s-length price doesn’t follow a single formula; instead, it requires establishing and weighing multiple factors such as economic circumstances, geography, market conditions, business strategies. When an AI driven system is trained on and captures even one less of all the relevant deterministic factors, it may systematically mischaracterize legitimate business arrangements.

Balance of power? In a dispute between a taxpayer’s position and an AI-assisted assessment by SARS, whose position is more likely to prevail in a court of law?

One might argue that the taxpayer’s position should prevail when they provide robust and well-substantiated evidence, given that the burden of proof generally rests with the taxpayer. However, practice does not always mirror theory.

If AI‑driven assessments aren’t properly governed, they could subtly shift the balance of power in transfer pricing disputes. Both taxpayers and tax authorities need to think carefully about how AI is used. As AI becomes more embedded in transfer pricing audits, it is important to ensure that it enhances, rather than undermines, fairness, transparency, and the proper application of the arm’s-length price.

Nigeria

In Nigeria, transfer pricing audits have historically been characterized by complexity and procedural challenges. As audit activities intensified over the years, the methods used to conduct these reviews also began to evolve. The Federal Inland Revenue Service, now the Nigeria Revenue Service, through its international tax department, introduced several process improvement measures aimed at reducing past inefficiencies. This prompted a gradual shift from manual procedures toward more technology-enabled approaches. Today, document sharing platforms, virtual meeting tools, voice recording technology, enhanced analytics, and other digital solutions play an increasingly important role in how transfer pricing audits are conducted.

A milestone in this digital transformation is the migration of transfer pricing filings, including declarations, disclosures, and country-by-country reporting notifications, from the E-TP platform to TaxPro Max and now Rev360. This signals the authorities’ intention to create a “one source of truth” for tax data, which should invariably improve tax data quality and consistency. Although there is currently no evidence that the revenue service has deployed an advanced AI driven transfer pricing audit system, recent developments by the Nigeria tax administration suggest that the foundational architecture is being put in place. With the rapid growth of AI globally, and its increasing use in other jurisdictions, the question is becoming less about if large scale AI deployment will occur, and more about when.

This makes it essential to consider the safeguards and governance measures Nigeria must establish to ensure that future transfer pricing audits remain fair, robust, and well-regulated once AI-enabled systems are introduced.

Key Considerations

Safeguarding taxpayer rights. As AI becomes more embedded in the transfer pricing audit process, protecting taxpayer rights must remain a central principle. Under current laws, taxpayers are obligated to justify economic substance, pricing rationale, and commercial justification for their cross-border intragroup transactions. Any AI‑enabled audit workflow should preserve this right by ensuring taxpayers retain the opportunity to challenge assumptions, provide clarifications, and present additional evidence.

Responsible AI use. For AI to meaningfully enhance transfer pricing audits without compromising fairness and avoiding AI errors, the Nigerian revenue service and other stakeholders must preserve the human‑in‑the‑loop principle. AI systems, no matter how advanced, are only as reliable as the human oversight applied to the data they consume and the outputs they generate. Human review is essential to ensure that AI‑driven insights are interpreted within the proper economic, legal, and commercial context, especially in a judgment‑intensive discipline like transfer pricing.

Data quality. Since AI outputs are only as reliable as the data used to train and develop them, the establishment of a consolidated, accurate, and high‑quality data environment is critical. Ongoing improvements to tax administration platforms such as the development of a modern administrative system like Rev360, demonstrate progress toward unified taxpayer data systems and form part of the foundational infrastructure required for AI‑supported transfer pricing analysis.

Further, clear data governance rules, covering data validation, retention, privacy, and security, must be put in place to steward AI-assisted audits.

Structured engagement and preventing algorithmic bias. To minimize unnecessary disputes, the revenue service should proactively introduce structured engagement mechanisms throughout the transfer pricing audit process. This may include pre audit dialogue sessions, where taxpayers are given the opportunity to contextualize flagged transactions or explain economic arrangements before formal adjustments are considered. Where AI driven outputs significantly reshape the understanding of a taxpayer’s business model, obtaining explicit taxpayer clarification or at least offering clear disclosures would align with responsible AI best practices.

Internal agreement or Standard Operating Procedures (SOPs) on AI supported audits. Transparency will be critical for building trust and ensuring fairness as the revenue service adopts AI supported audit methods. To achieve this, the revenue service should develop clear internal guidelines that define how transfer pricing risks are identified using AI, specific data indicators that trigger enhanced reviews, and the circumstances under which AI tools may be deployed. These principles can be formalized in an internal framework and communicated to relevant teams to promote consistency and reduce uneven treatment across sectors. Without strong safeguards, AI algorithms may unintentionally target certain industries, misinterpret unique business models, or place undue weight on outlier data potentially creating new challenges instead of resolving existing ones.

What Can Multinationals Do?

As technological advancements accelerate in both Nigeria and South Africa, multinational enterprises can’t afford to take a passive stance. The regulatory landscape is shifting rapidly toward systems in which technology, and increasingly, artificial intelligence and related technologies, play a central role. To remain compliant, minimize disputes, and proactively manage emerging risks, multinationals must, in addition to robust technical positions, begin preparing for a more digitized and analytically intensive audit environment. The following measures can help them stay ahead:

Advance Pricing Agreement, or APA, programs. As tax authorities adopt more sophisticated audit technologies, APAs are likely to become increasingly attractive. By agreeing on transfer pricing methodologies upfront, and doing so collaboratively with tax authorities, multinational enterprises can obtain greater certainty and reduce the risk of prolonged, resource‑draining audit disputes. Bilateral APAs in particular provide protection against double taxation, which becomes even more critical in an environment where AI‑driven adjustments may heighten the frequency of cross‑border disagreements.

Data integrity and documentation consistency. With the South African revenue service already seemingly approaching an era of smart audits and the Nigerian revenue service laying the foundation for similar capabilities, taxpayers must ensure their data is clean, consistent, and audit‑ready across all reporting layers. High‑quality, internally aligned information across financial statements, transfer pricing local and master files, country-by-country reports, value-added tax filings, and income tax returns significantly reduces non‑compliance risks. Maintaining this level of consistency also lowers the likelihood of being repeatedly flagged by AI systems due to avoidable mismatches or inconsistencies across reporting documents.

Improved defensibility. Substance over form remains a foundational transfer pricing principle. Tax authorities are required to look beyond written contracts to assess how transactions actually occur in practice. Multinationals should therefore ensure that their transfer pricing outcomes are fully aligned with actual operational substance. This includes maintaining robust functional analyses and keeping clear evidence that profits are allocated in line with the functions performed, assets employed, and risks assumed.

Digital transparency and proactive engagement. Both the South African and Nigerian revenue services have emphasized their commitment to taxpayer service and transparency in their respective vision and mission statements. This suggests that tax authorities are not without understanding of the challenges brought on by the digital era. Multinationals should therefore take a proactive approach when risk triggers arise. This includes engaging early with the authorities to explain situations where automated systems may misinterpret their business models and preparing clear reconciliations for any discrepancies that might appear suspicious. Internally, businesses should also establish rapid response mechanisms to address digital audit queries promptly and consistently. This may involve setting up cross-functional teams, maintaining up-to-date documentation repositories, and developing protocols for responding efficiently to audit issues.

Invest in technology. Investing in technology remains one of the most significant competitive gaps for multinationals. As tax authorities adopt AI and other advanced technologies, organizations that continue to rely on manual or fragmented processes quickly find themselves at a disadvantage. Manual systems often introduce avoidable errors, timing inconsistencies, version control problems, incomplete audit trails, and weak data, all of which increase the likelihood of being flagged by automated risk assessment tools. Investing in modern technology is therefore not merely a matter of operational efficiency; it is essential for audit readiness, defensibility, and long-term compliance resilience.

Takeaways

Transfer pricing audits are at a new phase, where traditional, subjective analysis is increasingly intersecting with technology. South Africa offers a preview of what AI enabled transfer pricing audits could look like in practice. At the same time, it causes legitimate taxpayer concerns about over reliance on algorithms in an area of tax that ultimately rests on fact specific judgments.

Rapid AI adoption could tip the balance in transfer pricing audits in ways that disadvantage taxpayers, but this will only be the case when appropriate safeguards aren’t put in place. Robust taxpayer protection policies, transparent AI governance, strong data‑integrity practices, and meaningful human oversight are essential to ensuring that digital transformation enhances rather than undermines fairness.

There appear to be several points of optimism in the use of AI for transfer pricing audits: speed, enhanced transparency, reduced administrative burden, better audit quality, improved efficiency. However, alongside these advantages are core principles that must not be compromised. Fairness, transparency, proper taxpayer engagement, and sound human judgment all remain fundamental to credible transfer pricing administration. If these principles guide the deployment of AI‑supported audits, African tax authorities can achieve a balanced model, one that leverages technology for efficiency and precision while preserving the judgment, context, and fairness at the heart of sound transfer pricing administration.

This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law, Bloomberg Tax, and Bloomberg Government, or its owners.

Read More in This Series:
Artificial Intelligence and the Future of Tax Law-Journal Series

Read More on BLaw:
In Focus: Artificial Intelligence (AI)

Author Information

Christian Wiesener is an associate director of Global Transfer Pricing Services group at KPMG South Africa. Barbara Mbaebie is a senior manager in the Global Transfer Pricing Services group, and Akaoma Oseleis a manager in the Tax Technology & Transformation group, at KPMG Nigeria.

Interested in writing? Review our author guidelines, and submit pitches to Insights@bloombergindustry.com.