Will AI Fix Climate Change or Make it Worse? The Critical Role of Law and SDG 16 in Guiding Technology for Justice

Aftab Haider

Zhang Hui

 

Abstract: AI would not fix climate change by default; without legal guardrails, it can worsen the problem. Our thesis is that SDG 16 (accountable, transparent institutions) must be the backbone of climate-aligned AI. Most studies highlight AI’s potential but fail to provide sufficient empirical evidence proving its effectiveness in climate action. We propose a Net-Climate-Benefit Test: no AI deployment should proceed unless independently verified lifecycle emissions reductions clearly exceed the system’s own footprint. Translating these normative principles into an actionable governance toolkit requires a coordinated regulatory approach. Foremost, it demands mandatory transparency mandates regarding energy consumption and emissions metrics during training and inference. This must be structurally reinforced by standardised MRV protocols for climate claims, carbon-budgeted procurement constraints, and independent auditing mechanisms capable of penalising greenwashing. Significantly, the architecture must integrate equity safeguards to guarantee that structural benefits are equitably directed toward high-risk communities. We apply this framework to case studies in energy, manufacturing, and agriculture to show where AI succeeds, where it fails, and the specific legal reforms needed to close the gap. By establishing a law-based, transparent, and measurable approach, this work transforms AI’s theoretical potential into concrete climate solutions, deploying AI equitably to advance sustainable development.