The boundary problem (BP) asks who ought to be included by democratic procedure. Initially articulated as a challenge for democratic citizenship, the BP is later reframed as a difficulty faced by the ‘democratiing AI' (DAI) agenda. By mapping different senses of DAI, this article aims to provide a resolution to the BP in the context of DAI. In particular, it notes that DAI shifts our attention away from a crucial factor to make ethical decisions around AI: the power disparities within organizations that put employees advancing ethical AI in a vulnerable position. In this context, I show that workplace democracy provides a more promising normative aspiration than the DAI agenda, although some elements of the latter, such as its emphasis on subjecting tech firms to greater democratic control, serve to mitigate certain problems that workplace democracy is unable to address, for example, when members of tech firms unanimously prioritise ethically concerning AI practices.
