BIT by BIT: Regulation of Highly Capable AI Through the Remodelling of Bilateral Investment Treaties

Abstract

This study addresses the pressing need for regulatory frameworks to govern highly capable Artificial Intelligence (AI) systems, which pose significant risks such as potential misuse to create or enhance biological threats. The research investigates the indirect role of Bilateral Investment Treaties (BITs) in regulating highly capable AI, particularly focusing on redesigning BIT clauses in light of the risks associated with the technology. Through qualitative research and examination of BIT clauses, the study finds that while current BITs do not explicitly address the risks posed by highly capable AI, there is potential for redesigning BIT provisions to do so by: i) amending the definition of ‘investment’ to explicitly include data centres, which play a significant role in the development of highly capable AI, and ii) broadening the scope of scenarios where police powers can be exercised to include AI risks. This study subsequently invites further research on the implementation of the proposed redesigned BITs to mitigate the risks associated with highly capable AI.