Will the Climate Pay the Price as the EU Strives to Become a World Leader in AI? : An analysis of how the EU incorporates climate commitments into its AI legislation

Sammanfattning

The adverse effects of climate change pose a threat that requires enhanced climate action on a global scale. Within the EU, this has been acknowledged through the adoption of the European Climate Law along with core environmental principles, such as the polluter pays principle. These measures aim to mitigate the environmental impacts within the EU, supporting the EU's objective of becoming a climate-neutral continent by 2050. In pursuit of this objective, the EU has undertaken several climate commitments. At the same time, the EU has presented an ambition to position itself as a global leader in AI by becoming the world's first AI continent. As part of this strategy, the EU has adopted the AI Act, seeking to promote the development and deployment of trustworthy AI within the Union. However, the AI Act does not explicitly address the environmental impacts associated with the increasing deployment of AI systems. Although the regulatory framework demonstrates awareness of potential risks by categorising AI systems according to their level of risk, environmental considerations are notably absent from the AI Act. Instead, AI is primarily framed as a tool capable of mitigating the adverse effects of climate change, rather than as a technology that generates significant environmental impacts. This is noteworthy given that the types of AI systems being implemented, especially the generative AI and GPAI models, have a significant impact on the environment. This is due to these systems being considerably more resource-intensive than earlier generations of AI models. In light of this, the EU must take such environmental considerations into account when developing a framework for AI systems. Despite this, the AI Act does not incorporate the objectives established in the European Climate Law in a way that would ensure the development and deployment of climate-friendly AI across the Union. This thesis therefore examines alternative approaches to AI regulation. In particular, it examines whether the adoption of a new act or directive would be a more appropriate regulatory response, taking into account considerations of innovation, global competitiveness and the practical implications for EU businesses. Accordingly, an AI Climate Directive is deemed to be the most appropriate solution, as such a framework would allow the EU to address the existing regulatory gaps by incorporating key environmental principles into AI deployment.