Towards Sustainable AI: Technical and Regulatory Challenges

Current AI regulations, like the AI Act and AI Liability frameworks, focus on trustworthiness and accountability but lack a robust approach to environmental sustainability. This talk addresses the AI sector's significant environmental footprint. It highlights the need for adapting existing regulations, including EU environmental law, GDPR, and the AI Act, to better account for sustainability. Proposals include transparency mechanisms for AI systems' GHG emissions and a regulatory toolkit with co-regulation, sustainability-by-design principles, and consumption caps, potentially integrated into the EU Emissions Trading Scheme. This framework aims to ensure sustainable AI and offers a model for other high-emission technologies, tackling the twin major transformations of AI and climate change mitigation.