Proportionate Risk Management for Tax-Related AI Systems: Analysis and Implications under EU and US Law

This article analyses how emerging EU and US regulatory frameworks on artificial intelligence (AI) shape the operation of tax administrations. On the EU side, it examines how the General Data Protection Regulation (GDPR) and the AI Act jointly constrain Member State tax authorities, highlighting the tension between rights-based protections, such as data subject rights and limits on automated decision-making, and the Act’s risk-based obligations for high-risk AI systems used in tax enforcement. For the United States, the article reviews recent public-sector AI initiatives affecting the IRS, focusing on how evolving federal guidance and the repeal of the Safe and Trustworthy AI Executive Order can create uncertainty around applicable standards for tax-related AI (TAI) systems.The comparative analysis shows convergence in regulatory aims but fragmentation in implementation. To address the need to set a uniform framework of protection and the deficiencies of both approaches, the article proposes a proportionality-based risk management framework tailored to specific AI use cases in tax administration, offering structured, scalable safeguards for both authorities and taxpayers.