The AI Governance Gap in Financial Advice: What AFSL Holders Need to Know
Published 22 September 2026
Short answer: 74% of Australian advice practices already use or plan to use AI (Adviser Ratings, 2025 Australian Financial Advice Landscape Report), but governance hasn't kept pace. When ASIC reviewed 23 licensees using AI, only 12 had AI governance policies addressing fairness and bias (ASIC REP 798, October 2024). ASIC hasn't issued AI-specific rules, but existing licensee obligations apply in full, whatever tools a practice uses.
Most financial advice practices are already using AI in some form, or about to. Far fewer have decided, in writing, how it should be used. That gap between adoption and governance is a genuine, largely invisible exposure sitting inside a lot of AFSL holders' compliance files right now.
The numbers
- Adoption: Adviser Ratings' 2025 Australian Financial Advice Landscape Report found 74% of Australian advice practices already use or plan to use AI.
- Governance: ASIC's REP 798, Beware the gap: Governance arrangements in the face of AI innovation (29 October 2024), reviewed 23 licensees using AI. Only 12 had AI governance policies that addressed fairness and bias. ASIC's own title for the report says it plainly: AI use is moving faster than the governance around it.
No AI-specific rule doesn't mean no rule
Unlike accounting (TPB(GS) 55/2026) or law (QLS Guidance Statement No. 37), there's no single named ASIC instrument spelling out how AI use interacts with a licensee's obligations. That absence is sometimes read as "nothing to worry about yet." It's closer to the opposite. ASIC has been clear that existing licensee accountability applies in full regardless of the tools used, including for cyber security and third-party data handling.
A few things follow from that:
- Accountability sits with the licensee, not the vendor. A reputable AI product doesn't move responsibility off your compliance file.
- There's no carve-out for embedded features. AI already switched on inside CRM, research, or portfolio tools counts the same as a deliberate AI session.
- A written policy is the practical starting point, even without an AI-specific rule to build it against.
Why the gap matters more than it looks
A governance gap isn't just paperwork. It means that when AI is used inside a practice (a paraplanner summarising a client's financial position, an adviser drafting correspondence, a research tool pulling in portfolio detail), nobody has actually decided what's acceptable, what client data can go where, or who checks the output.
If something goes wrong, whether a data incident, a complaint, or an ASIC inquiry, "we didn't have a policy" is not a strong starting position, with or without an AI-specific instrument.
What's at stake for the client relationship
Financial advice runs on a specific kind of disclosure: net worth, family structure, sometimes information a client hasn't told their own family. Clients extend that trust because they believe the relationship is discreet. In a client's mind, "trust me with your money" and "trust me with your data" are the same promise. A paraplanner pasting a client's portfolio into a public AI tool doesn't just create a compliance question. It puts that discretion in doubt.
Closing the gap, in two parts
- Write the policy. Document which tools are approved, what client data can and can't be entered, and who reviews AI-assisted output before it reaches a client. That closes most of the exposure on paper and gives you something concrete if ASIC or a client ever asks.
- Shrink what the policy has to cover. A policy is easier to write, and easier to follow, when the answer to "where does client data go" is simple. Running AI on hardware inside your own office means there's no chain of external vendors, sub-processors, or retention terms to write a policy around, because there isn't one.
The practices most exposed right now aren't the ones avoiding AI. They're the ones already using it without having decided, on paper, what that means.
Common questions
Has ASIC issued specific rules on AI use in financial advice?
Not as a dedicated instrument. ASIC has been clear that existing licensee obligations apply in full regardless of which tools a practice uses.
How many advice practices use AI?
74% of Australian advice practices already use or plan to use AI, according to Adviser Ratings' 2025 Australian Financial Advice Landscape Report.
Do most licensees have AI governance in place?
Not consistently. In ASIC's REP 798 (October 2024), only 12 of 23 licensees reviewed had AI governance policies addressing fairness and bias.
Do we need to stop using AI while we write a policy?
No. It means documenting which tools are approved, what client data can go into them, and who reviews the output.
Sources: Adviser Ratings, 2025 Australian Financial Advice Landscape Report; ASIC, REP 798, Beware the gap: Governance arrangements in the face of AI innovation (29 October 2024).