<p>This article develops the concept of the last labour frontier to analyse uneven artificial intelligence adoption as a contested struggle over the centrality or peripheralisation of human labour. Existing debates on technological unemployment and automation exposure identify tasks and occupations at risk, but they often understate the social repositioning of human work when algorithmic systems become infrastructures of production, coordination and evaluation. The article argues that AI adoption does not merely eliminate jobs; it redefines human labour as central, hybrid, residual, peripheral or excluded. This frontier is shaped through conflict among employers, technology firms, workers, trade unions, professions and states over whether labour remains a productive and social institution or becomes replaceable, supervisory and institutionally contained. The argument is supported by evidence on uneven occupational exposure to generative AI, changing skill demand, platform-work regulation, and AI governance in the European Union. The contribution is to shift AI-and-work debates from the measurement of displacement to the contested social positioning of labour, showing that the future of AI-mediated work is an industrial relations and institutional struggle over whether human labour remains central, becomes hybrid, is reduced to supervision, is peripheralised or is excluded.</p>