A special, sector-specific tax on AI justified by the arguments that it is built on the public's collective knowledge or that it runs on public infrastructure is the wrong instrument: both justifications lack a limiting principle and apply to every firm and prior technology, the legitimate grievances underneath them are better served by copyright markets and litigation (for uncompensated training data) and by the existing progressive corporate and capital-gains base (for concentrated profits), and a dedicated AI levy would fall on buyers and workers through prices and wages, entrench incumbents against startups and open source, and tax a productivity input at the moment economies most need it; the larger real tax distortion runs the other way, since the current code over-subsidises automation relative to labour.

Steelman anchors (the strongest pro-tax case, represented before rebuttal): Bill Gates' robot tax (Quartz, Feb 2017, the worker-vs-robot $50,000 parity argument); Andrew Yang's data-as-property 'data dividend' (CNBC, 17 Oct 2019) and California Governor Newsom's 'Data Dividend for Californians' (State of the State, Feb 2019); Mariana Mazzucato's The Entrepreneurial State (2013) on the public origins of the internet, GPS, touchscreen, and voice assistant (DARPA / Defense Department / NASA); and the 2026 currency, OpenAI's policy blueprint proposing a shift of the tax base toward corporate income and capital gains plus levies on automated labour, with Altman and Khosla's under-$100,000 income-tax-exemption proposal (Fortune, 7 Apr 2026). Counter-case anchors: the European Parliament adopted its civil-law robotics resolution but rejected the robot tax inside it on 16 Feb 2017, resolution carried 396-123-85 (European Parliament A8-0005/2017 / TA-8-2017-0051); Lawrence Summers, 'Robots are wealth creators and taxing them is illogical' (Washington Post op-ed, 5 Mar 2017): no logic to singling out robots versus other labour-displacing tools, 'protectionism against progress', and an administrability problem; the licensing market and the courts already route training-data compensation to rights-holders (News Corp-OpenAI up to $250m over five years, the largest content-licensing deal in publishing history; Financial Times $5-10m a year; Axel Springer about $13m a year, per Digiday's 2024 deal timeline; NYT v OpenAI core copyright claims allowed to proceed toward trial, NPR, 26 Mar 2025, with no fair-use summary judgment expected before summer 2026); Acemoglu, Manera and Restrepo, 'Does the US Tax Code Favor Automation?' (Brookings Papers on Economic Activity, 2020): effective labour tax above 28.5% versus roughly 5% on equipment-and-software capital drives excessive automation, and moving to neutral taxation would raise employment by about 4%; Guerreiro, Rebelo and Teles, 'Should Robots Be Taxed?' (Review of Economic Studies, 2022): even the canonical pro-tax model finds the optimal robot tax small, temporary, and falling to zero as today's routine workers retire. Scope: this is a claim about the right policy instrument, not a claim that AI firms should pay no tax (the position is that AI profits should be taxed as any profits are) and not a prediction about any single vendor. Reviewable prediction embedded: no major Western jurisdiction enacts a special, sector-specific AI or robot tax (as distinct from applying general taxes or copyright remedies) before the next review. 90-day review cadence (27 Aug 2026). Triggers to revisit before cadence: (a) a major jurisdiction (EU, US, UK) introduces a serious bill for a dedicated AI / compute / automation tax, which would move the empirical prediction toward Partial; (b) a binding fair-use ruling in NYT v OpenAI or a peer case that closes the copyright-market route, which would strengthen the case for a statutory levy and warrant re-examination; (c) adoption of a national data-dividend statute that pays individuals for training-data use, which would partially realise the first pillar and warrant a move toward Partial. Siblings: the EU AI Act regulatory-readiness pieces (the compliance-cost mechanism the article points buyers toward) and the AI-and-jobs task-level coverage at /how-ai-changes-jobs-task-level-frame/.