What ICAEW, ACCA, AAT and ICAS have published on AI

The common basis

The ICAEW, ACCA, AAT and ICAS codes of ethics are based on the code of the International Ethics Standards Board for Accountants (IESBA), with its five fundamental principles: integrity, objectivity, professional competence and due care, confidentiality and professional behaviour (ICAEW: generative AI and ethics; AAT: professional ethics).

IESBA added technology-related material to its code. ICAEW's guide says the code was updated in 2022 to set out threats such as undue reliance on technology (section 112.1) and limited understanding of technology (113.1 A2) (ICAEW: generative AI and ethics). ICAS says the technology-related revisions to the ICAS Code of Ethics became effective on 1 January 2025 (ICAS, 14 July 2026).

Guidance by body

The PCRT topical guidance on AI

ICAS reported on 19 January 2026 that the seven PCRT bodies had published topical guidance on the ethical use of AI tools for tax work (ICAS, 19 January 2026). AAT links the same guidance from its professional ethics page. It is the most detailed of the bodies' publications and it applies to members in practice.

It works through each fundamental principle with risks and safeguards. Under integrity it suggests a statement in the engagement letter that AI-enabled software may be used, and an audit trail of prompts and sources. Under objectivity it describes automation bias, the tendency to favour output from automated systems even when other information casts doubt on it. Under professional competence and due care it requires staff training on any tool used and asks for output to be reviewed as if prepared by a junior colleague. Under confidentiality it says putting client data into public tools is likely to breach confidentiality without client consent (PCRT AI topical guidance).

ICAEW

ICAEW's generative AI guide applies each principle to AI. On professional competence it says generative AI tools do not make up for a lack of professional knowledge and skill. On confidentiality it says confidential information should not be loaded into public generative AI tools, because there is limited visibility of who it is shared with, how it is secured and how long it is kept. It also says accountants should ask suppliers the right questions and get evidence about training data and how bias is managed (ICAEW: generative AI and ethics).

Its 7 September 2026 article on validating AI content adds a working rule: use AI as a starting point and check every technical claim against a trusted source (ICAEW, 7 September 2026).

ACCA

ACCA's position paper of August 2023 puts accountability and AI literacy at the centre: professionals need to understand the capabilities, limits and uses of AI in their own field. It lists the risks of explainability, bias, privacy and security, legal uncertainty, inaccuracy and the magnifying effect of one error repeated across large volumes of data (ACCA: AI in the finance profession, August 2023).

The ACCA and CISI report of October 2025 says the IESBA technology revisions took effect in December 2024. Its suggestions for practitioners include disclosing AI use, validating tools, avoiding unapproved or public tools for client data without authorisation, and due diligence on suppliers covering audit rights, deletion and where data is held (ACCA and CISI, October 2025).

AAT

AAT's Code of Professional Ethics is based on the IESBA code and sets out the same five principles. Section 140 of the code covers confidentiality, including the duty to take reasonable steps so that staff and those giving advice or assistance respect it (AAT Code of Professional Ethics). For AI, AAT points members to the PCRT topical guidance (AAT: professional ethics).

ICAS

ICAS reported IESBA's snapshot on the ethical implications of technology, including AI, on 14 July 2026, and IESBA's characteristics-based guidance on emerging technologies on 20 August 2026. The second says professional accountants remain responsible for their judgements and decisions when technology is used (ICAS, 14 July 2026; ICAS, 20 August 2026).

What the guidance asks of a practice

• Understand any tool before relying on it, and train the staff who use it.
• Review output with professional scepticism, and be able to explain how a conclusion was reached.
• Keep confidential client data out of tools the practice does not control, unless the client has agreed.
• Tell clients that AI tools may be used, through the engagement letter.
• Keep a record of the tools used and of the review.
• Ask suppliers for evidence about data handling, training data and security.