Over the last few years, there has been a growing recognition that AI safety efforts must increasingly account for the realities, needs, and aspirations of the Global South, in order to ensure that AI governance frameworks are equitable and representative. Given the prominence of meaningful inclusion of the Global South and Global South perspectives in AI safety processes, this report offers insights from the direction and experiences of existing AI Safety Institutes (AISIs) for upcoming Global South AISIs. 

The analysis under this report is supported by a mapping of existing and upcoming AISIs in the UK, the US, the EU, Japan, Singapore, France, Canada, India, Brazil, Kenya, and China  on the basis of elements including governance structure, objectives, and functions. 

Experiences from existing AISIs may not be fully replicable for jurisdictions in the Global South which are currently in the process of establishing their AISIs. Existing global trends like innovation-centric shifts away from safety in AI governance focus areas could potentially have repercussions for the Global South, where unique local and contextual risks like bias, and low levels of digital literacy persist. Jurisdictions in the Global South may also have limited access or challenges in allocation of resources arising from economic and structural asymmetries. In comparison to the Global North, there is also limited representation of the Global South in AI safety processes, which can impact both the functioning of AISIs themselves, as well as the partnerships they may potentially enter. 

Through our report, we have identified certain key insights relating to priorities, functions and partnerships of AISIs in the Global South. There is a need for Global South AISIs to account for opportunities and risks prevailing in existing AISIs and the AI ecosystem and adopt measures contextualised to their local and regional realities. This objective can be achieved by establishing meaningful multistakeholder governance structures, increasing representation of the Global South in AI safety processes, and strategically harnessing and allocating technical and financial resources. 

This report is part of a larger project on the evolving landscape of AI Safety in the Global South, supported by Konrad Adenauer Stiftung (KAS).