Pollution

Mitigation strategy

1. **Mandate the Optimization of AI Models and Infrastructure for Resource Efficiency** Implement stringent performance standards requiring the use of energy-efficient AI algorithms (e.g., model quantization, smaller language models) and specialized, low-power hardware. Simultaneously, compel the transition of all AI-serving data centers and computational facilities to 100% verifiable low-carbon or renewable energy sources to directly reduce air pollution from greenhouse gas emissions. 2. **Establish and Enforce Standardized Environmental Impact Measurement and Disclosure** Develop and ratify global, standardized protocols for the comprehensive life-cycle assessment (LCA) and mandatory public disclosure of the environmental footprints—including energy consumption, total carbon emissions, and water withdrawal—associated with AI systems. This is necessary to facilitate effective governance and ensure accountability across the AI value chain. 3. **Implement Circular Economy Strategies for AI Hardware and Water Management** Require the adoption of sustainable procurement policies for AI hardware, focusing on materials with lower embodied emissions and minimal reliance on conflict minerals. Furthermore, enforce the widespread deployment of water-efficient cooling technologies (e.g., submersion or evaporative cooling) and closed-loop systems to minimize water withdrawal and mitigate the pollution risks associated with electronic waste and hardware disposal.