Lloyd’s Register Foundation and Careful Industries announce forthcoming Foresight Review on the Safe Adoption of AI

Lloyd’s Register Foundation has today announced the start of a new Foresight review on the safe adoption of AI. The review will enable evidence-based decision making for AI developers, purchasers, implementers, and policymakers of the potential harms to people, infrastructure and the environment that may occur through AI adoption.

Guided by input from global experts with lived, learnt, and practical experience of AI risks, the project will focus on unfolding and future scenarios for AI adoption in safety critical areas. This will widen the scope of assurance for both policymakers and AI developers and provide an accessible mechanism for assessing the “real-world” impacts of AI development. 

To achieve this, the project will focus on current and emergent impacts of AI adoption in three areas:

Worker safety 

• Analysing the impacts on workers of developing and deploying AI systems, from data labelling and moderation labour to the risk factors involved in working alongside robotics and cyber-physical systems.

Environmental safety

• Exploring the impacts of data-centres and mining of rare earth metals for AI infratructure, juxtaposed with the considerable contributions of AI and machine learning towards environmental monitoring and renewable energy. 

Infrastructure resilience

• Evaluating the vulnerabilities caused by rapid development and roll-out of AI systems across critical infrastructure systems, and the governance and resilience risks associated with AI adoption..

Speaking on this new collaboration, Rachel Coldicutt OBE, Founder and Executive Director of Careful Industries and Author of the Foresight Review, said: “AI is a complex field of computer science in which 'safety' is often viewed as a technical development issue, assessed via pre-deployment quality assurance checks. However, few AI systems are truly finished at deployment: new outcomes and impacts on people and systems can emerge throughout their lifecycle and – as these emergent technologies are increasingly integrated into critical infrastructures – they create many new challenges for engineering safety and assurance.

“We’re delighted to be conducting Lloyd's Register Foundation’s latest Foresight Review, and I’m confident the recommendations will help stakeholders across both technological and infrastructural domains make informed decisions about AI development and deployment.”

The Foresight Review process will identify a range of potential harms related to the integration of AI that can be applied to different safety-critical contexts by technical and non-technical experts. The findings will emphasise the importance of whole-lifecycle approaches to assurance for policymakers, developers and purchasers of AI products across multiple domains.

Reflecting on the need for foresight in this area, Jan Przydatek, Director of Technologies at Lloyd’s Register Foundation, said: “The scale of current and predicted AI uptake across infrastructure assets, systems, and industrial processes is accelerating rapidly. This uptake will inevitably influence the safety of people that interact with these systems and will add new forms of complexity to these systems. As this complexity develops at pace, we need to be aware of the risks to physical, digital and environmental safety so that these risks can be managed towards safer outcomes.

“As a global safety charity, one of our priorities is to ensure future AI systems, whatever shape they take, are deployed in a safe and reliable manner and where possible, enhancing safety. The foresight review will demystify the subject of AI, and its findings will be action-oriented for industry leaders and policymakers around the world as they take crucial steps towards the scaled-up deployment of AI in safety-critical systems.”

The project will combine desk research with qualitative and participatory research and foresight methods in five key locations: India, South Africa, Kenya, China and the UK. Throughout, collaboration with experts with lived, learnt, and practical experience will be central to identifying a range of unfolding and emergent issues. 

The Foresight Review is expected to launch in June 2026.