Special Issue Editorial: Extraction by Design—AI, Value, and the Future of Work

Abstract

In August 2025, Australia’s Productivity Commission acknowledged that “large AI models are already being trained on copyrighted materials without consent or compensation” [1, p. 25]. In doing so, they cited evidence already documented by numerous scholars and advocates regarding how traditional expectations about safeguarding the ownership of completed work could no longer be assured. Yet, shocking authors and artists, the commission continued onwards to propose that the legal concept of “fair use” can be interpreted as allowing tech companies to freely use copyrighted material to train artificial intelligence (AI) models even without explicit consent [2]. Even while raising the possibility that this concession for tech companies would come at the expense of copyright owners, the commission acknowledged that the local economies of AI innovation might still feel no benefit whatsoever: “At present, large AI models are trained overseas, not in Australia. It is unclear whether the introduction of a [text and data mining] TDM exception would change this trend” [1, p. 28]. It is easy to see the injustice of giving away the rights of authors and artists to their work and (meagre) livelihoods, especially when doing so doesn’t even promise direct payoffs back to Australian workers. But how can we move beyond this initial critical reaction towards also contextualizing such transfers of intellectual property (IP) within a more robust understanding of how work and its products are currently governed in the age of AI?