Explainability as a Constitutional Limit on Algorithmic Tax Power: A Comparative Analysis

Abstract

The wide application of algorithmic systems in tax administration has reshaped the exercise of tax power and posed unprecedented challenges to traditional constitutional constraints on taxation. Classical constitutional principles, including legality, equality and due process, were formulated for human decision-makers and prove structurally insufficient to address the opacity, probabilistic logic and systemic biases of algorithmic tax decisions. This article argues that explainability should be established as a new structural constitutional limit on algorithmic tax power, rather than merely a transparency tool or ethical guideline. It distinguishes explainability from transparency, accountability and interpretability, and clarifies its core function as a substitute for human intentionality, which guarantees taxpayers’ right to contestation and judicial review. Based on a comparative analysis of the regulatory frameworks in the European Union, the United States, China and Brazil, this study examines respective governance models and existing deficits. It finds that all jurisdictions lack specialized tax laws targeting algorithmic decision-making. To safeguard taxpayers’ fundamental rights, explainability must be combined with contestability, auditability and effective human oversight. This research offers a new theoretical framework for adapting constitutional tax law to the era of artificial intelligence.