As a business owner or investor, time is tight. It makes sense that people ask AI tools for quick answers on deductions, super contributions, or “best structure” ideas. The responses sound confident, arrive instantly, and feel low-risk.
The risk is real.
Australia’s tax and super rules are detailed, fact-specific, and change over time. AI can help you understand the basics, it can also send you down the wrong path, and the clean-up often costs more than getting advice early.
Where AI can help
AI is useful for learning and organising your thinking. For example, it can:
- explain terms like negative gearing, trust distributions, or concessional vs non-concessional contributions
- summarise what a rule is trying to do
- suggest a checklist of records to keep
- help you prepare questions for your accountant
Used this way, AI saves time because you arrive to the conversation clearer on the concepts.
Where AI falls over
Problems start when AI shifts from explaining to recommending what you should do.
Tax and super outcomes depend on your exact circumstances, your income, entity structure, residency, timing, asset history, future plans, and purpose. AI doesn’t know those facts unless you feed them in, and feeding personal financial details into public tools creates its own risks.
Even with good prompts, AI can’t apply professional judgement. It won’t weigh up competing risks, identify what matters most, or spot when a “clever idea” is actually a red flag.
The accuracy risk, confident but wrong
AI tools can generate responses that sound authoritative but are incomplete, outdated, or plain incorrect. That can look like:
- deductions that don’t apply to your situation
- capital gains outcomes that ignore key integrity rules
- super strategies that breach contribution caps or eligibility rules
- references to rulings, cases, or concessions that are wrong or don’t exist
A real example, the Tribunal calling it out
In Smith and Commissioner of Taxation [2026] ARTA 25, the Administrative Review Tribunal dealt directly with the issue of unverified AI-style legal research. The Tribunal warned that if AI is used, every case identified needs to be located and read to confirm it exists and supports what is being claimed, otherwise the Tribunal’s “time and scarce resources are being wasted, as the Tribunal must look for cases that do not exist and read cases that have no relevance at all”.
The ATO expects you to verify information
The ATO is not “anti-AI”. It uses analytics and technology internally. For taxpayers, the ATO’s public guidance warns that AI tools can produce false or inaccurate information, and the practical message is to verify what you rely on before acting.
If something is wrong, the ATO can amend the return, charge interest, and apply penalties. “The AI told me” won’t protect you.
Superannuation is high stakes
Super rules leave little margin for error, especially with SMSFs. Common problem areas include eligibility, timing, purpose tests, related-party dealings, and investment restrictions. A wrong move can mean non-compliance, forced unwind steps, and penalties that quickly add up.
Privacy and data security
There’s also a straightforward operational risk. If you enter personal or business financial details into an AI platform, you may lose control over where that data goes and how it is stored. That creates privacy and fraud exposure that often outweighs any time saved.
A smarter approach we recommend at Fortis – use AI as a learning tool, and keep it at that level.
For decisions that affect your tax position, cash flow, or long-term wealth, bring the question to us early. We can validate the idea against your full circumstances, flag risks you may not see, and confirm what records you need to support the outcome.
Bottom line, AI can be a helpful assistant. It isn’t your accountant.
If you have questions or concerns, please do not hesitate to contact our office to speak to one of our team.
