This article discusses the possibility of the rehabilitation of tax incentives to address the negative effects of artificial intelligence (AI) automation, especially since investments in new technologies resulting in automation are largely subsidized. This idea is premised on the need to restore the neutrality of the tax system, understood in both its equity and efficiency dimensions. The article uses the concept of tax expenditures as the theoretical background for evaluating automation tax incentives and their potential limitation as to whether it might advance the tax system’s normative goals. By examining several automation tax preferences in light of tax neutrality, some risks were identified for both equity and efficiency based on the workers-machines substitution effect. The article concludes that a tax policy reducing tax incentives on AI-driven automation would be proportional to the aim pursued if the scope of the measure is designed by referring to the substitution effect between robots and workers and would not be such as to impede investment in AI that is socially desirable.