Ai agent decisions complicate income tax fairness and policy
Welfare-Opaque Income: Taxation under AI-Agent Delegation
Computers and Society
Summary
When AI agents make economic decisions based on hidden rules, it becomes harder for governments to understand how income changes relate to people's well-being. The authors call this situation 'welfare-opaque income' because the same income changes can mean different things for social welfare, depending on hidden AI decision processes. They develop a way to adjust tax policies by including information about how AI executes decisions, which helps government better estimate the effects of tax changes. They also simulate AI behaviors across different systems to see how these hidden rules impact income and taxation.
What this means in practice
- •For tax policymakers: Design tax systems that incorporate hidden AI decision behaviors to better predict welfare outcomes and optimize tax rates.
- •For economic modelers: Improve simulations of taxation effects by including AI agent execution decision data to refine welfare impact estimates.
Authors
Yukun Zhang, Kemu Xu, Yishen Chen
Abstract
We study income taxation when an AI agent implements economically relevant choices through a rule hidden from the government. Alongside unobserved productive ability, this hidden preference-to-execution mapping creates \emph{double unobservability}: the same observable tax-base response can carry different welfare consequences. We call the resulting income \emph{welfare-opaque}. Our constructions show that tax-base statistics can coincide while reform welfare effects differ, even when mechanical welfare weights are identical. We derive an optimal-tax condition that adds a response-weighted execution wedge to the familiar sufficient statistics. A higher marginal rate gains a corrective benefit under local over-execution and an additional cost under local under-execution. Observing the wedge identifies the welfare effect of a marginal reform at the prevailing schedule; bounds on it deliver bounds on that effect. A controlled laboratory compares 4,500 model runs across five AI engines. Faithful delegation selects the score maximizer in essentially all runs. Conflicted objectives produce heterogeneous responses: Claude largely preserves the score maximizer, GLM moves predominantly downward, and GPT-mini and Qwen show concentrated lower-tail increases. Qwen also makes substantial downward adjustments. Different engines locate their departures at different points and in different directions of the designed distribution. Explicit scores align model rankings; formula-based objective instructions yield more uneven agreement. Qwen shows a clear positive tax-by-objective interaction, but its direction does not generalize across engines and the pooled sign depends on its inclusion. The analysis identifies execution information as a complement to conventional tax-base statistics.