AI versus Product Safety & Environmental Footprint

Nigel Burtt, Environmental Regulation Consultant – IAMT

“I’m sorry Dave, I’m afraid I can’t do that” intones the HAL9000 computer in Stanley Kubrick’s seminal 1968 movie ‘2001: A Space Odyssey’ to astronaut Dave Bowman after the computer discovers the spaceship crew believes it has malfunctioned and needs to be disabled. But had the artificial intelligence of the computer failed, or did it just consider that it is superior to the humans on the ship? (“This mission is too important for me to allow you to jeopardize it.”) Or can it foresee the ending of the story and realised that human evolution is about to make a huge advance, potentially making computers obsolete? Of course, this is all fiction and although the timeline is incorrect, Kubrick’s vision of the artificial intelligence, or AI, at the heart of Arthur C. Clarke’s story is incredibly prescient.

As AI becomes embedded in the products we design, how do we deal with the risks that software and functionality updates that could be ‘self-configured’ by the product itself may conflict with its originally intended safe operation and it and its necessary infrastructure do not cause danger to human health and our environment? These are questions that politicians and regulators are trying to address without the full knowledge of what is and is not possible and practicable.

This topic was discussed at the most recent UK’s Office for Product Safety & Standards (OPSS) Business Reference Panel meeting in London on Monday 1^st June as there is an ongoing UK government consultation on a package of proposed reforms of the UK product safety framework including aspects of AI. This consultation builds upon last year’s AI Opportunities Action Plan and the creation of the AI Security Institute (AISI) which is a research organisation within the UK government’s Department for Science, Innovation and Technology (DSIT.) The UK’s AISI is just one institute in an international network of safety institutes that was agreed in 2024 at a Summit in South Korea. For example, the network includes the European Union’s EAIO and the United States’ CAISI. It is interesting to note, however, that both the UK and USA decided in 2025 to re-brand their institutes to remove the word ‘Safety’ from the original names of the organisations.

For now, in the UK it is intended that the 2025 Product Regulation and Metrology Act (PRAM) will be used to regulate the use of AI, where it may affect safety, when it is an operational component of a physical product, that is an AI-Enabled Product (AIEP). However, the government also recognises that AI is also becoming integral to the product lifecycle, influencing design, manufacturing, and supply chain aspects. Clearly these influences also have an impact on compliance with regards to testing, standards, certification and market surveillance to ensure quality and safety.

The UK government notes that AI is already transforming industries by driving innovation and enhancing user experiences across society and that AI is increasingly being built into consumer products. However, as yet, it feels that current evidence shows no widespread safety issues linked to AI-enabled consumer products. It is seeking feedback from stakeholders on any real-world experiences with AIEPs, including the risks they pose and how the UK can effectively protect business and consumers without stifling innovation.

Alongside these matters is the difficult subject of the impact of AI on the environmental footprint of businesses and products. Sally Radwan, the Chief Digital Officer of the United Nations Environment Programme (UNEP) has been quoted to say that, “Governments are racing to develop national AI strategies, but rarely do they take the environment and sustainability into account. The lack of environmental guardrails is no less dangerous than the lack of other AI-related safeguards.” In 2024, UNEP produced a report entitled, “Artificial intelligence (AI) end-to-end: The environmental impact of the full AI life cycle needs to be comprehensively assessed”  but, so far, the issues raised in this document still do not seem to have a high priority. It recommended, for example, that countries should establish standardized procedures for measuring the environmental impact of AI and develop regulations that require companies to disclose the direct environmental consequences of AIEPs and services.

The environmental aspects highlighted by UNEP certainly don’t appear directly in the UK’s consultation process for reform of the product safety and supply chain duty regulations with regards to AIEPs as referred to above, although there perhaps is an opportunity offered to respond along these lines within questions A26 “What do you think are the current or potential harms associated with AI-enabled products?” and A27 “How can we ensure that the reformed product safety framework effectively addresses the unique challenges posed by AI-enabled products and digital innovations, while supporting innovation?”