Regulate or Revise: Addressing Algorithmic Bias in AI-driven Residential Mortgage Underwriting in Australia

Australian banks have recently indicated their intention to offer fully automated digital mortgages, driven by artificial intelligence (AI). While this will likely improve loan approval times, it risks causing algorithmic bias. Utilising a hypothetical -based methodology, this article analyses how the Racial Discrimination Act 1975 (Cth), National Consumer Credit Protection Act 2009 (Cth) and Corporations Act 2001 (Cth) currently address algorithmic bias in AI -driven residential mortgage underwriting. Identifying legislative gaps, this article then turns to what type of law reform is best fit to address this algorithmic bias. This article evaluates two typical reform approaches, technology regulation and revising anti -discrimination laws, before turning to a more novel approach, revising financial regulation laws. This article argues that the latter approach is most favourable as it empowers the Australian Securities and Investments Commission to protect against the main causes of algorithmic bias: biased data, poor system design and testing, and opaque machine learning models.