Towards sustainable AI: The prospects for energy and environmental law in policy and regulation

AI operates through a material infrastructure of data centres, electricity supply, cooling systems, water use, land and transmission capacity, whose legal significance has received insufficient attention. Among others, rising electricity demand, growing water consumption and carbon consequences linked to the energy mix suggest that AI can create problems which fall within environmental and energy law. The article shows that AI specific regulation leaves much of this infrastructural burden insufficiently addressed and evaluates the potential for environmental and energy law to provide a distinct route for regulating AI by shaping the conditions under which AI systems are trained and deployed. In doing so, this article examines the energy and environmental challenges associated with AI, explains the limitations of AI specific regulation and considers existing approaches in the EU, China and the United States to developing energy and environmental regulation relevant to AI. Finally, this article makes practical policy recommendations for more comprehensively incorporating energy and environmental law as central considerations within policy and regulation relevant to AI with a view to promoting sustainable AI.