この記事は現在英語でのみご利用いただけます。
AI in Australian Superannuation: Trustee Obligations, Member Communications, and Investment AI
Superannuation funds are using AI in member communications, investment management, complaints handling, and fraud detection. The SIS Act trustee obligations, APRA prudential standards, and ASIC conduct requirements create a governance framework that most funds have not fully mapped.
Key Takeaways
The SIS Act sole purpose test (section 62) constrains how superannuation funds can use member data — AI applications that generate revenue for the fund beyond member benefit purposes may create compliance issues.
Trustee obligations require that AI in investment decisions reflects the trustee's informed judgment — AI can inform but not replace trustee decision-making. Documented oversight of investment AI is essential.
APRA's superannuation prudential standards (SPS 230, SPS 220) apply to AI systems used in material business activities, including member communications AI, investment systems, and complaints handling.
ASIC's regulatory guidance on financial product advice applies to AI that provides retirement savings advice or guidance — the best interests duty applies even when delivered through an AI system.
Member communications AI must comply with RG 271 internal dispute resolution requirements — AI-generated responses to member complaints must meet the same standards as human responses.
"情報提供のみを目的としています。この記事は法律、規制、財務または専門的なアドバイスを構成するものではありません。具体的なアドバイスについては、資格を持つ専門家にご相談ください。"
Superannuation AI: the unique governance context
Australian superannuation funds hold approximately $3.5 trillion in retirement savings on behalf of 16 million Australians. The governance obligations that apply to this capital — and the AI systems that increasingly help manage it — are correspondingly significant. Superannuation trustees operate under the SIS Act, APRA's superannuation prudential standards, ASIC's financial services conduct requirements, and the Privacy Act. AI systems used in superannuation must be assessed against all of these simultaneously.
The unique feature of superannuation governance is the trustee duty: trustees are required to act in the best financial interests of members. AI systems must be assessed not just for technical performance but for consistency with this duty. An AI system that optimises for fund revenue rather than member outcomes, or that provides efficiency gains but introduces risks that affect member retirement savings, is not consistent with trustee obligations.
The sole purpose test and member data AI
Section 62 of the SIS Act requires that a superannuation fund is maintained solely for prescribed core and ancillary purposes — primarily the provision of retirement benefits to members. This constraint applies to how the fund uses member data. AI applications that use member data for purposes beyond member benefit — for example, selling data-derived insights to third parties, using member profiles for insurance cross-selling beyond member benefit — may not be consistent with the sole purpose test.
Before deploying any AI system that processes member data for commercial purposes beyond direct member benefit, trustees should obtain specific legal advice on sole purpose test compliance. This is an area where the interaction between AI capabilities and superannuation law creates genuine novel risk.
Investment AI and trustee decision-making
Many superannuation funds use AI in investment management — for portfolio optimisation, risk monitoring, factor investing, and ESG analysis. The trustee obligation creates a specific governance requirement for investment AI: trustees must exercise their own informed judgment, not simply implement AI recommendations. This means trustees must understand, at a sufficient level to exercise judgment, what the AI investment systems are doing, what their limitations are, and why they are making particular recommendations.
The documentation requirement is significant: trustee minutes should record that investment decisions — including decisions to adopt or maintain AI-driven investment approaches — reflect the trustees' informed judgment. "The algorithm recommended it" is not adequate documentation of trustee decision-making.
Member communications AI and the advice boundary
Superannuation funds are increasingly using AI in member communications — answering questions about accounts, projecting retirement outcomes, explaining investment options, and responding to complaints. This creates a significant compliance risk: AI systems that provide personalised projections, comparisons, or recommendations about a member's superannuation may cross the boundary from general information into financial product advice, triggering Australian Financial Services Licence requirements and the best interests duty.
Funds should clearly define what AI member communications systems can and cannot say, and ensure there are escalation pathways to qualified financial advisers for members who need advice. The ASIC distinction between general information, general advice, and personal advice applies fully to AI-delivered communications.