APES 110 and AI: the Code of Ethics, applied to AI tools.

APES 110 Code of Ethics for Professional Accountants (including Independence Standards) is the ethical code for Australian accountants. It is issued by the Accounting Professional & Ethical Standards Board (APESB), based on the international code published by the International Federation of Accountants, and every member of Chartered Accountants Australia and New Zealand, CPA Australia and the Institute of Public Accountants is required to comply with it. This page explains how the Code is built, what its five fundamental principles require, and how each applies when a firm uses AI, with a detailed guide on each obligation.

General information only. This guide summarises published professional standards and legislation and is not legal or professional advice. Read the primary sources linked throughout, and take advice on your firm’s specific position.

How the Code is built

Part 1 applies to every member. It sets out the five fundamental principles in Section 110 and the conceptual framework in Section 120 that members use to apply them. Part 2 adds material for members in business, Part 3 for members in public practice, Parts 4A and 4B set the independence standards for audit, review and other assurance work, and Part 5 covers sustainability assurance. Most AI questions in an accounting firm are answered in Parts 1 and 3.

Technology is already in the Code

APES 110 does not use the words “artificial intelligence”. It does address technology directly, most of it through technology-related revisions APESB issued in June 2024, effective from 1 January 2025 for Parts 1 to 3. These are the provisions an accounting firm’s AI use runs into:

  • Objectivity (R112.1) must not be compromised by bias, conflicts of interest, or undue influence of, or undue reliance on, individuals, organisations, technology or other factors.
  • Professional competence (113.1 A3) requires a continuing awareness and understanding of technology-related developments relevant to the work.
  • Confidentiality (114.1 A1) covers how information is collected, used, transferred, stored or retained, disseminated and destroyed, and the Code addresses authorisation to use client information in developing products or technology (114.3 A3).
  • Using the output of technology (R320.11) requires a member in public practice to determine whether using a technology’s output is appropriate for the intended purpose.
  • Bias (120.12 A2): the Code’s examples of bias to be aware of include automation bias, the tendency to favour output from automated systems even when other information calls it into question.

The five fundamental principles, applied to AI

Section 110 sets out five principles. Each has an AI dimension, and most have a detailed guide:

  • Integrity (Subsection 111). Members must not knowingly be associated with information that contains materially false or misleading statements, or statements provided recklessly. Passing on AI output nobody has checked is the obvious way to end up associated with exactly that, which is why the guide to checking AI output matters here too.
  • Objectivity (Subsection 112). Judgement must not be compromised by bias or by undue reliance on technology. The conceptual framework guide works through automation bias and the technology threats the Code names.
  • Professional competence and due care (Subsection 113). Understanding the tools you rely on, checking what they produce, and supervising the people who use them.
  • Confidentiality (Subsection 114). The principle most AI decisions turn on, because it governs where client information goes.
  • Professional behaviour (Subsection 115). Comply with relevant laws, including the Privacy Act and the TFN Rule, and make no exaggerated claims in marketing, which includes how a firm describes its own AI-assisted services.

The obligations that sit alongside APES 110

APES 110 binds members of the professional bodies. Registered tax and BAS agents also answer to the Tax Practitioners Board, and any practice holding individual clients’ tax file numbers is bound by the TFN Rule. Each has a guide:

  • The TPB Code of Professional Conduct. Statutory confidentiality, competence and reasonable care obligations under the Tax Agent Services Act 2009, and the TPB’s supervision and quality management expectations.
  • TPB(GS) 55/2026, the Board’s AI guidance. Issued 22 July 2026: the practitioner stays accountable for accuracy, AI output must be verified and documented, and client permission is needed before client information enters an AI tool, because an AI provider is a third party.
  • Tax file numbers. The TFN Rule limits how TFN information is used and disclosed and who may access it, which bears directly on what may go into an AI tool.
  • The firm AI policy. How a written policy turns all of the above into rules staff can follow at deadline.

A practical evaluation checklist

Name the data

List what would actually enter the tool: TFNs, financials, identity documents, correspondence. If the answer in practice is "whatever staff paste", the policy is not real.

Trace the flow

Where is it processed, where is it stored, who can access it, under what law? Get vendor answers in writing and keep them with your engagement files.

Test against Subsection 114 and the TPB Code

Could you explain to a client, or the Board, why this handling respects confidentiality? If the explanation depends on trusting an offshore vendor’s internal controls, decide whether that is a position you want to hold.

Test against R320.11

Is relying on this tool’s output appropriate for the purpose? Who in the firm understands it well enough to say so, and how was it tested?

Prefer structure over policy

Rules staff must remember fail at deadline. Architectures where the compliant path is also the productive one, such as processing that never leaves the practice, fail less.

Where deployment posture fits

Most of what APES 110 asks is about member behaviour: checking output, supervising staff, exercising judgement. No tool does that for you. Confidentiality is different, because it turns on where client information travels, and that is an architecture question. When processing happens on hardware inside the firm, client information is never disclosed to an AI provider, which is why Ermos builds on-premise AI for accounting practices. If the firm already lives in Microsoft 365, our Copilot comparison works through the data-flow fine print. Neither APES 110 nor the TPB Code prohibits AI. Both make the firm accountable for where client information goes.

Common questions

What is APES 110?
APES 110 Code of Ethics for Professional Accountants (including Independence Standards) is the ethical code for Australian accountants, issued by the Accounting Professional & Ethical Standards Board. It is based on the international code published by the International Federation of Accountants, and members of CA ANZ, CPA Australia and the IPA are required to comply with it. Its core is five fundamental principles: integrity, objectivity, professional competence and due care, confidentiality, and professional behaviour.
Does APES 110 mention AI?
It does not use the term. It does address technology directly: undue reliance on technology as a threat to objectivity, technology-related developments as part of competence, and, since the technology-related revisions took effect on 1 January 2025, a requirement for members in public practice to determine whether using the output of technology is appropriate for its purpose. It also lists automation bias among the examples of bias to be aware of.
Can accountants use ChatGPT with client data?
APES 110 Subsection 114 prohibits disclosing confidential client information unless there is a legal or professional duty or right to do so, or the client has authorised it. The TPB's AI guidance, TPB(GS) 55/2026, says practitioners must obtain the client's permission before entering client information into AI tools, because a third party is any entity other than the client and the practitioner. Without that permission and a documented review of the tool, the cleaner answer is to keep client data out of it, or to use AI that runs inside the practice.
Do we need client consent to use AI tools?
If client information will be disclosed to a third-party service, client authorisation is the most defensible footing, through engagement terms or specific consent. APES 110 suggests that when authorisation is sought to use information in developing technology, the firm communicates, preferably in writing, the information involved, the purpose and who will carry out the activity. If processing stays inside the practice, there is no third-party disclosure to authorise.

See Dominion answering from your own documents.

A demonstration takes 30 minutes, uses no client data, and comes with no obligation.

Book a demo