Algorithms and Trustworthiness in Tax Administration

Abstract

This article examines the integration of algorithms into the work of Tax Administrations. It argues that these tools are not neutral instruments but reflect historical biases and institutional choices. While they offer opportunities for greater efficiency and consistency in areas such as fraud detection and service provision, their use also raises complex legal, ethical, and governance challenges.

The authors explore how algorithms, particularly those based on machine learning and artificial intelligence, reshape decision-making processes. They show that although these technologies can implement efficient and trusted methods, their use as digital civil servants requires careful oversight. Through practical examples, such as VAT fraud detection and AI-assisted taxpayer guidance, the paper highlights both the potential and the risks involved.

Tax professionals, the authors argue, must play a central role in defining the objectives, assessing the limitations, and ensuring the ethical use of algorithmic tools. Algorithms should serve as support systems, not replacements for legal reasoning or institutional judgment.

The paper concludes that the challenge is not whether to adopt algorithms, but how to govern their use responsibly, balancing innovation with the duty to uphold public trust, fairness, and accountability.