Generative AI regulation in the UK and Australia: comparing two national attempts at un-regulation

In AI regulation, much literature focusses on the so-called ‘Brussels Effect’ [8], the copying of legislation and standards from the European Union (EU), where the Regulation on AI was passed in August 2024 [1, 3,9], often on the basis companies are already compliant with that regime. Literature also examines the lesser known ‘Beijing Effect’, the adoption of Chinese style “command and control” law by less liberal-democratic nations especially in the neighbourhood of China in East Asia [2]. There is also a ‘Washington Effect’, whereby other nations copy the less regulatory approach of US federal law, which occurred in Internet liability in the 1990s [12] but may be less likely under Trump Administration moving towards protectionism rather than ‘unregulation’. In this paper, we focus on the regulation of AI by two advanced compute nations that have adopted neither approach despite their proximity to these exemplars. In AI, both the UK and Australia have adopted a cautious regulatory approach to AI, focussed on empowering sectoral regulators to adopt advanced market knowledge and mapping exercises, a ‘light touch’ without specific legislation [5,6]. We argue that both nations are adopting an ‘unregulation’ or sectoral regulation approach which is differentiated from that in Washington, Brussels or Beijing.