As your trusted accountant, I’ve been watching closely how artificial intelligence (AI) is increasingly used across business functions — including bookkeeping, payroll, tax-planning and financial reporting. AI offers tremendous promise: faster processing, smarter insights, reduced routine labour. But a recent high-profile mis-step by a Big 4 accounting firm shows exactly how things can go wrong — and why that matters for you, especially if you’re a small or medium business owner relying on accounting and payroll advice.
In July 2025, a Big 4 accounting firm published a report for a government department which reportedly incorporated generative-AI tools. The final version was found to contain fabricated references, mis-attributed court judgments and other factual errors — prompting Deloitte to issue a refund on part of its fee.
What’s significant is not just the mistakes themselves, but the system failures behind them — weak review, over-reliance on AI drafts, lack of provenance and human oversight.
For you as a small business owner, or someone managing payroll and accounts, this isn’t just consulting-world drama. It’s a reminder: when AI enters accounting, payroll or compliance workflows, it must be handled with discipline or it becomes a risk, not simply a productivity gain.
- What went wrong – and why you should care
For many small businesses, the core issues here translate into real-world risks:
- The AI-generated elements introduced hallucinations (i.e., convincing but false content) — references and quotes that didn’t exist.
- The review process apparently failed to catch those errors — meaning that the work reaching the client (in this case the government) was compromised.
- Deloitte’s error undermines trust; in our sector (accounting, payroll) trust equals compliance, assurance and credibility. If your accounting tool or process uses AI in a weak way, you expose yourself to flawed advice, misclassification of payroll, incorrect tax treatment or audit issues.
In short: AI isn’t a magic bullet. Just using it doesn’t guarantee accuracy or reliability. Like we always say in accounting: checks, reviews, documented rationale still matter.
- Six AI-risks every SME in accounting and payroll should heed
While your business might not be drafting government-reports, here are risks that do apply in the SME context:
- Hallucinated outputs – AI might suggest incorrect tax rulings, mis-code payroll categories, invent vendor names or dates.
- Data privacy & confidentiality – When staff paste sensitive payroll, tax or client-data into generic chatbots, there’s risk of leakage or misuse.
- Compliance drift – The law, regulation (ATO, Fair Work, Superannuation, Payroll Tax) change regularly. If AI isn’t aligned to the correct Australian rules, you might act on outdated or incorrect advice.
- Opaque AI vendor behaviour – Some software may leverage “AI assistance” without telling you how much, or what validation occurs.
- Over-automation without human oversight – If you let AI generate payroll changes, journals, reconciliations without a human review step, small errors become big issues.
- Accountability gaps – If an AI-tool produces guidance but no one takes ownership, then when something goes wrong, who is accountable? This is especially important for SMEs with lean teams.
These are the kind of failures exposed in this case — and we should use this as a wake-up call for our own workflows.
- A practical AI governance checklist for your accounting & payroll
How can you use AI safely — and still gain benefit? As your accounting partner, here’s what I recommend we implement (or check) together:
- Define approved tools & data-use policy: Use only tools that meet Australian privacy standards, avoid pasting TFNs, bank details, personal data into unvetted chatbots.
- Human-in-the-loop review: Every AI-assisted output (e.g., payroll suggestion, cost-centre coding, financial forecast) must be reviewed by a qualified human (you or me).
- Source-of-truth rule: Any advice or output from AI must be checked against legislation, ATO rulings, Fair Work awards, or professional judgement — AI doesn’t replace that.
- Traceability / audit-trail: Keep records of what was generated by AI, prompts used, who reviewed, what changes were made. Good for compliance and future proofing.
- Vendor due diligence: If your accounting or payroll software says “AI assisted”, check how the model was validated, how often it’s updated for Australian tax/award changes, what error-rate exists.
- Training & awareness for your team: Staff who use AI tools need to know the risks — that the tool may be wrong, that they must apply professional judgement.
- Start small, monitor results: Use AI for low-risk tasks (first draft letters, supplier-coding suggestions) before trusting it with payroll changes or tax advice.
When these guardrails are in place, AI becomes a booster for your accounting and payroll efficiency — not a liability.
- Where AI does work well in SME finance
It’s not all risk — there are practical win-wins for small business:
- Invoice-processing-tool that suggests accounting codes (human reviews afterwards).
- Spend-analysis dashboards that flag anomalies (you still decide on action).
- Drafting standard letters or memos (you still personalise and check).
- Payroll-software that highlights unusual payroll entries (you still approve).
In these cases you get improved efficiency and better oversight — but the key is controlled use, not delegate-and-forget.
- Book a Consultation: Stay Ahead of AI Risks
If you’re unsure how AI might impact your accounting or payroll, let’s talk. The Deloitte case shows what happens when AI goes unchecked — and small businesses can’t afford those mistakes.
Book a quick consultation with us to:
- Understand the real risks of AI in your finance systems
- Learn how to protect your data, compliance, and payroll accuracy
- Get practical steps to use AI safely and confidently
AI should work for your business, not against it.
Book your consultation today and let’s make sure your accounting and payroll stay secure and future-ready.
Conclusion
This case is more than just a headline — it’s a reminder that in accounting and payroll, precision, accountability and trust matter. AI can absolutely be part of your business toolkit, but only when used with professional rigour. Let’s ensure your finance processes are efficient, modern and safe.
