POLICY AND JUSTICE DIMENSIONS OF ARTIFICIAL INTELLIGENCE WATER CONSUMPTION: TRANSPARENCY, REGULATION, AND ENVIRONMENTAL EQUITY

Abstract

The rapid expansion of artificial intelligence (AI) has resulted in significant but mostly unnoticed costs to the environment. In this paper, the policy and justice aspects of the water footprint of AI are analyzed. Corporate disclosures of Google, Microsoft and Amazon show major differences in transparency: Google discloses information on the facility level, Microsoft withholds location-specific breakdowns, and Amazon presents only efficiency measures, which obscures the total footprint, estimated as exceeding global bottled water consumption. An overview of regulatory initiatives such as the U.S. Artificial Intelligence Environmental Impacts Act of 2024, the EU AI Act, and proposed Planning and Infrastructure Act amendments in the UK reveals a fragmented and underdeveloped policy landscape. The recent U.S. Supreme Court case Loper Bright Enterprises v. Raimondo complicates the authority of federal agencies, meaning Congress will need more specific legislation. This paper argues that mandatory disclosure, standardized reporting distinguishing Scope-1 and Scope-2 consumption, and binding water efficiency targets are necessary to ensure accountability and prevent disproportionate burdens on vulnerable communities