Drawing on Goodin and Spiekermann (2018)’s epistemic theory of democracy, this paper frames the rapid proliferation of generative AI as a profound epistemic crisis. If democratic efficacy relies on collective truth tracking, deepfakes fracture this shared reality and fundamentally erode electoral integrity. In response, Australian jurisdictions have begun legislating against distinct forms of digital manipulation, though these laws differ in their target and scope.
South Australia enacted laws criminalising harmful and sexually degrading deepfakes. This paper argues that this establishes a critical jurisdictional baseline for addressing AI driven harms. In contrast, New South Wales explicitly targeted electoral integrity by passing the Electoral Legislation Amendment (Elections) Act 2026. This Act targets political deepfakes in the electoral context. Meanwhile, a comprehensive federal approach remains unresolved. Although parallel federal proposals have circulated, such as Senator David Pocock’s 2025 push against deepfakes created without consent, this paper contends that Australia currently lacks a unified national framework to protect its elections from disinformation generated by AI.
This paper analyses Australia’s emerging patchwork regulatory response. It evaluates the efficacy of localised bans in a borderless digital information environment and examines a core constitutional tension highlighted in NSW parliamentary research: balancing the need for epistemic security with the implied freedom of political communication. In conclusion, this paper argues that without a cohesive federal strategy, jurisdictional interventions may prove insufficient to mitigate the "liar’s dividend", leaving Australian voters vulnerable to algorithmic manipulation.
