This exploratory note is intended to map out the functioning of Large Language Models (LLMs), which form the basis of some of the popular generative AI services, against key data protection requirements globally.

The note takes a law-agnostic approach and maps out the functioning of LLMs against conceptual elements which are common to most data protection regulations.

Given that this is a rapidly evolving space, this document is inherently dynamic and future iterations may outline other risks and include recommendations for addressing the same.