AI in Financial Advice: Innovate Fast, but Quality will always be Forefront

Australian advice practices have gone all-in on AI. Adoption has surged from under half of practices to roughly three-quarters in the space of a year and most of the rest say they're piloting or planning to follow. The appeal is obvious - faster drafting and less time buried in admin. But there's a quieter number sitting underneath that growth curve: a clear majority of practices using AI still don't have a formal policy governing how it's used. That gap is where real risk lives.

The Regulator's Position is Simple - and Unforgiving.

Don't expect a dedicated "AI rulebook" from ASIC. The message instead is that the existing framework is technology-neutral: your obligations to act in best interest and avoid misleading conduct apply exactly the same way whether a Statement of Advice was typed by a human or drafted by a language model.

In practice, that means using AI doesn't shift accountability - if an AI tool gets a recommendation wrong, the licensee wears it, not the vendor.

So, How Do You Use AI the Right Way?

1. Tier your use cases by risk.

Not every AI application carries the same exposure. Marketing copy and scheduling sit at the low-risk end, client engagement emails and review summaries sit in the middle and drafting advice documents, generating recommendations, or touching product selection sit firmly in high-risk territory and deserve the tightest controls.

2. Fold AI into your existing compliance program - don't bolt it on.

If AI is helping draft an SoA, your compliance review process needs to explicitly say so and explicitly cover it. The same goes for AML/CTF programs: regulators increasingly expect documentation of how any AI models involved are trained, validated and calibrated - meaning no space for simply trusting the tech.

3. Watch where client data actually goes.

Free, consumer-grade AI tools are an easy way to breach the Australian Privacy Principles as they do not offer any data protection - upload client info to a free version of ChatGPT and you’ve put yourself and your client at risk. Before adopting any tool, check for Australian hosting or contractual privacy commitments and run a privacy impact assessment if the tool touches personal information in any meaningful way.

4. Keep a human firmly in the loop - AI isn’t meant to take your job!

AI should accelerate your expertise, not replace your judgement. Treat outputs as a trigger for manual review, not a shortcut past one.

5. Document everything.

Policy, training, escalation paths, vendor due diligence - if it isn't written down, it's hard to demonstrate later that you were managing the risk rather than just hoping for the best.

The Bigger Shift Advisers Can't Ignore

Plenty of consumers have already stopped waiting for the industry to solve the advice-access problem - they're opening a chatbot and asking their financial questions directly. The direction the industry seems to be heading in response is that AI-powered guidance should ideally sit under an AFSL, be transparent about its limits, be grounded in someone's real circumstances rather than generic assumptions and be accountable when it's wrong, but it is not there yet.

For advice practices, that's both a warning and an opportunity. The firms that get hurt here won't be the ones using AI - it'll be the ones using it without policy, oversight and anyone able to explain how a given output was produced. The firms that get ahead will treat AI as a regulated tool from day one, not an experiment they can clean up later.

Move Fast on Adoption, But Move Just as Fast on Governance.

In this environment, "we didn't think AI counted" won't be a defence anyone wants to test in front of ASIC.

If it feels like more effort than its worth in a small-scale business, consulting can help take the weight off your desk and make sure you are ahead of the game. Solara Consulting Services can help with technology review, implementation and governance so you get the best of all worlds.

‍