top of page
J Huang and S Pettigrove

ASIC finds AI governance failings in financial services


As the financial services industry embraces artificial intelligence (AI) in the delivery of products and services, a new report from the Australian Securities and Investments Commission (ASIC) highlights a critical governance gap that could have significant implications for both firms and consumers.


The report, titled "Beware the gap: Governance arrangements in the face of AI innovation," (the Report) provides a comprehensive review of 624 "AI use cases" by 23 Australian financial services and credit licensees, how they identify and mitigate associated consumer risks, and their governance arrangements.


Accelerating AI Adoption


The report reveals that AI adoption among licensees is accelerating rapidly, with a notable increase in the use of more complex and opaque techniques such as generative AI. Most AI use cases are relatively new, with many being less than two years old or still in development. ASIC's findings are summarised in the figure below, noting that the Report only collected data up to December 2023.



In the case of Generative AI, the Report says it is often used to generate first drafts of materials, or responses to customers in carefully constrained circumstances. Some other notable use cases include:


  • supervised learning which is mostly used to predict if a consumer is likely to take out a financial product using explainable models such as logistic regression, or used to derive prices, rates or forecast future series; and

  • deep learning models that were mostly used for natural language processing and optical character recognition, primarily when scanning analogue form data to speed up loan, insurance, or other form-heavy business processes.


This rapid adoption underscores the transformative potential of AI in the financial services sector, offering opportunities for enhanced efficiency, improved customer experience, and innovative product offerings.


Governance and risk management gaps


While the Report finds that the way licensees are using AI is mostly cautious - that models were not providing ungoverned outputs or replacing human judgement - it also identifies significant gaps in how licensees manage AI risks, particularly those unique to AI, such as algorithmic bias.


Governance arrangements vary widely among licensees, with some lagging behind in updating their risk management frameworks to reflect the evolving risks and challenges posed by AI. This disparity in governance practices raises concerns about the potential for consumer harm if AI use outpaces the development of adequate governance frameworks.


Consumer Impact


The Report says AI has the potential to amplify existing risks to consumers and introduce new ones, such as bias, discrimination, and the provision of false information. The Report warns that without robust governance arrangements, the rapid adoption of AI could lead to significant consumer harm. This highlights the need for financial services firms to prioritise the development and implementation of comprehensive AI governance frameworks that address these risks.


Recommendations for Licensees


To mitigate these risks, the Report provides several recommendations for licensees:


  1. Update governance arrangements: Ensure that governance frameworks lead AI use and are regularly updated to reflect the evolving nature of AI.

  2. Consider AI strategy and ethics: Reflect on questions related to AI strategy, ethics, accountability, risk management, and oversight.

  3. Enhance risk management: Develop robust risk management practices that address the unique risks posed by AI, including algorithmic bias.


Looking Ahead


As AI increasingly reshapes the financial services landscape, it is imperative for firms to strike a balance between innovation and consumer protection. By proactively addressing the governance gaps identified in the ASIC report, financial services firms can help harness the benefits of AI while safeguarding consumer interests.


For potential clients and industry professionals, staying informed about these developments and understanding the importance of robust AI governance will be crucial in navigating the evolving regulatory landscape and maintaining trust in the financial services sector.


Written by J Huang and S Pettigrove

Commenti


bottom of page