This paper proposes the Taxpayer Displacement Levy (TDL), a fiscal mechanism designed for an AI economy in which profitable firms may substitute artificial intelligence systems for human taxpayers. The central claim is not that AI should be taxed as technology, nor that productivity gains should be punished. The claim is narrower: when a financially strong enterprise deploys AI in a way that removes a worker from the tax base, and when the resulting loss in labor-side public revenue is not offset by corporate-side tax recovery or by the worker's expected return to an equivalent tax position, the remaining fiscal gap constitutes a public externality that may be subject to a targeted levy. The paper defines workers as public revenue interfaces: not merely private cost items on a corporate income statement, but fiscal infrastructure through which income tax, payroll tax, social insurance contributions, consumption tax, and local public revenue flow into the state. AI labor substitution therefore removes not only wages but a bundle of public revenue streams attached to the worker as a fiscal node. The paper distinguishes TDL from traditional robot taxes: robot taxes target machines; TDL targets displacement behavior. Robot taxes struggle with software AI; TDL applies to any technology that removes a taxpayer from the fiscal circuit. Robot taxes risk penalizing innovation; TDL is triggered only by financially non-necessary, AI-related labor substitution that generates an uninternalized fiscal gap. The proposed audit equation has both a single-period and a dynamic form. In single-period form: TDL Taxable Gap = First-order Labor-side Tax Revenue Loss − Incremental Corporate-side Tax Recovery − Expected Tax Base Restoration within the Re-taxation Window. In dynamic form: TDL Taxable Gap = Σt δt(Lt − Ct − Rt). The three terms have different epistemic status: the first is an accounting-observable quantity, the second a tax-model estimate, and the third a probability-distribution expectation. The procedure does not require false precision. It requires disclosure of assumptions, data, and uncertainty intervals. The paper grounds this framework in the precautionary principle, current tax-code capital bias, the 2025 OBBBA changes to bonus depreciation and domestic research expensing, the Falk-Tsoukalas AI layoff trap model, and the IRS paradox: a tax authority itself reducing personnel while expanding AI use, with large estimated revenue consequences. S3S4S5S6S7
Chen et al. (Wed,) studied this question.