London, 3 March 2025 – Guidance setting out how to measure an organization’s AI carbon footprint, and the main environmental sustainability factors that should be considered when using AI has been published by BSI, just weeks after the UK Government set out its aspiration to mitigate the sustainability risks of AI in the AI Opportunities Action Plan.
While there is widespread optimism that AI can help solve societal challenges, a study last year found that generative AI might use some 33 times more energy than machines running task-specific software. With organizations globally scrambling to invest in AI tools, the Technical Report (TR) sets out how to manage and measure sustainable AI.
Environmentally Sustainable Artificial Intelligence (PD CEN/CLC TR 18145:2025) is designed to provide guidance to AI developers and deployers - as well as executives and decision-makers - with explanations and techniques to support low-carbon technologies for AI, particularly in the context of available European energy sources. It includes information on energy resourcing and infrastructure, and specifically the processes and physical systems involved in the production, distribution, and consumption of energy, as well as how to precisely measure an organization’s AI carbon footprint and provide footprint data.
The focus on the European AI landscape and European-specific aspects of AI demands for resources, comes as last year 20% of the UK’s electricity was imported directly from the EU.
By providing standardised comparison metrics, the TR’s users will be able to quantitively assess the value and efficiency of their AI implementation. The intention is for this to help developers and deployers measure productivity gains, cost efficiency and overall performance compared to more conventional technologies, to help inform executive decisions about AI implementation.
The guidance includes:
- standardized terms related to sustainable AI and machine learning
- methods for software, hardware and location-based solutions for the reduction of energy consumption
- information on water use, pollution, hardware and data centres
- policy for lifecycle considerations and the circular economy
- use cases for energy-saving, optimizing natural resources, and managing climate change and sustainable ICT networks
- formulae to measure AI carbon footprints
Mark Thirlwell, Global Digital Director at BSI said: “The impact of AI on the fight against climate change is nuanced. AI is playing an increasingly central role in our lives and in the way organizations operate. Its ability to digest and respond to information at pace offers enormous potential to be a critical tool in our armoury, to enhance sustainability and support our pathway to net zero. At the same time, AI is extremely resource hungry. It is key to consider the amount of energy used in processing the data that sits behind AI.
“BSI's publication of this Technical Report is designed to guide organizations to balance technological advancement with environmental responsibility, by providing essential tools and insights to help measure and reduce the carbon footprint of their AI systems. By aligning AI with energy-efficient, low-carbon technologies, we can collaborate to accelerate progress towards an innovative future and a sustainable world."
Scott Steedman, Director-General, Standards at BSI said: “As AI continues to evolve, so must our approach to managing its environmental impact. The release of this Technical Report is very timely, providing organizations with the knowledge not only to leverage AI for growth but to do so with sustainability at the core of their strategies. It's not just about reducing carbon emissions, it’s about developing an AI ecosystem that is itself sustainable.”
Download the Technical Report today by visiting:
https://knowledge.bsigroup.com/products/environmentally-sustainable-artificial-intelligence
