The UK and US have recently signed a memorandum of understanding to collaborate on developing tests for advanced artificial intelligence models [3], with a specific focus on AI safety.


The UK-US AI Safety Institutes will work together to align scientific approaches and accelerate the creation of evaluation suites for AI models. This partnership aims to address safety risks associated with the rapid deployment of powerful AI systems and ensure the safe development of AI for societal benefit [3]. Plans include joint testing exercises on publicly accessible models [2], sharing capabilities and information on AI capabilities and risks, personnel exchanges [2] [5], and technical research on AI safety and security [2] [4]. The collaboration follows a commitment made at the AI Safety Summit [3], where major AI firms agreed to a voluntary scheme for evaluating and testing new AI models before release [1]. The Department for Science [1], Innovation and Technology (DSIT) stated that the partnership with the US will help governments stay ahead of emerging AI risks [1], with plans for similar collaborations with other countries in the future [1]. Both countries view the safe development of AI as a global issue and believe that collaboration is essential to harnessing its potential while mitigating risks to national security and society [3]. The US and UK AI Safety Institutes will conduct joint testing exercises on public models and exchange expert personnel to enhance shared capabilities [4]. Both countries will share information on AI systems and models [4], as well as conduct technical research on AI security and safety [4]. This collaboration aims to improve understanding of AI systems [4], conduct more robust evaluations [4], and issue more rigorous guidance to ensure AI safety now and in the future [4].


The collaboration between the UK and US on AI safety is crucial for addressing safety risks associated with powerful AI systems and ensuring the safe development of AI for societal benefit. By working together [2] [3], both countries can stay ahead of emerging AI risks and enhance their shared capabilities. This partnership sets a precedent for future collaborations with other countries and underscores the global importance of AI safety. The joint efforts of the US and UK AI Safety Institutes will lead to more robust evaluations and guidance to ensure AI safety now and in the future.