Does Professional Indemnity Insurance Cover AI Mistakes?

Every registered tax and BAS agent in Australia carries professional indemnity insurance because the law requires it — the Tax Practitioners Board makes adequate PI cover a condition of registration, and agents must review it at least once a year at renewal. Lawyers, financial advisers and most consultants carry it too, and no government tender gets signed without it. Here is the question almost nobody has put to their broker: if an AI tool drafts an error that reaches a client, does that policy actually pay the claim?

Right now the honest answer is 'it depends' — and that ambiguity is the exposure. The efficiency AI brings to advice letters, file notes and compliance summaries is real. The insurance that sits behind that work was, in almost every case, written before AI was in the workflow at all.

Australian courts and regulators have already settled who is liable when AI invents a fact: the firm that put its name to the work, not the AI vendor. We covered that when an AI-assisted report cost Deloitte a $97,000 refund and the firm — not the software — wore it. More than twenty Australian court matters since 2024 have turned on AI-fabricated citations, and the through-line is always the same: 'the AI got it wrong' is not a defence.

Professional indemnity is the backstop for exactly that kind of claim — a professional-negligence allegation arising from your work product. So the practical risk isn't only that AI produces an invisible error. It's that when a client acts on that error and comes after you, the policy you rely on to make it survivable may respond in a way you didn't expect.

Insurers have a term for this: 'silent AI' — cover that neither clearly includes nor clearly excludes an AI-driven loss, because the wording predates the technology. That silence is now being closed, but on the insurer's terms. In its Risk Radar 2026, Australian insurance law firm Sparke Helmore reports insurers are developing specialised 'affirmative' policies to address AI-specific risks such as model drift — which tells you where the market is heading, and that the default policy today is ambiguous rather than reassuring.

The professional-liability specialists say the same thing to firms: act now. Global insurance law firm DAC Beachcroft's 2026 professional-liability predictions urge firms to develop internal AI policies, train staff, and review their PI cover for gaps — specifically how a policy would respond to multiple errors caused by a single technology failure. Even the insurance channel is catching up: a NIBA-surveyed preparedness gap reported by Insurance Business in August 2026 found 83% of brokers expect automation to have a significant impact, but only 61% feel prepared for it. Don't assume your broker has already sorted this.

Want to know what this means for your business? We'll map it in one call.

This is an admin leverage problem before it is an insurance one. The Tax Practitioners Board requires registered tax and BAS agents to hold adequate PI cover from an APRA-approved insurer — a minimum of $1 million in aggregate cover for practitioners turning over more than $500,000 — and to review it every year at renewal. That annual review is the natural checkpoint. It is also the moment an underwriter is most likely to ask how you use and govern AI, because your answer now shapes their risk.

The firms that renew cleanly will be the ones that can show, on paper, where AI sits in their work and who checks it. The ones that can't are betting that a silent policy resolves in their favour after a claim — a bad bet in any year, a worse one as affirmative AI wordings arrive. If you're not sure where AI already touches your client work, start by mapping your own exposure before a renewal questionnaire — or a claim — does it for you.

Three moves, none of them expensive. First, build a one-page AI register — every tool that touches client work, what it does, and what data it sees. Second, write down your verification gate: a named person checks AI-assisted output against source before it leaves the firm, the same discipline that keeps you out of the liability cases, and exactly the evidence an insurer wants to see. Third, put two questions to your broker in writing: how does our current policy respond to a loss caused by an AI tool, and is affirmative AI cover available to us? The answers tell you whether you're carrying a gap. For the full regulatory picture, our guide to Australia's AI compliance rules maps how liability, privacy and disclosure fit together.

Key takeaways

Common questions

Related notes

Not sure your firm's AI is on the right side of your cover?

Book a call and we'll map where AI sits in your work, document the verification gate insurers now look for, and get you renewal-ready — so an AI error doesn't turn into an uninsured one.