This is a plain-English summary of published guidance, not legal advice.
What ICAEW has published on AI
- Generative AI guide: ethics. Applies the five fundamental principles to AI. Don't put confidential client information into public generative AI tools, watch for automation bias, and remember the tools are “not fully qualified accountants”.
- Code of Ethics technology provisions (from 1 July 2025). They apply to all technologies. Keep up with technology relevant to your work, judge whether relying on a technology's output is reasonable, don't become overdependent on it, and protect confidential information in any form.
- PCRT topical guidance on AI tools (19 January 2026). For tax work. Review AI output “as if it were prepared by a less experienced junior colleague”, consider mentioning AI in the engagement letter, and don't put client data into public tools without consent.
- AI and client data: complying with the principle of confidentiality (November 2026). Points firms to subsection 114 of the Code, calls AI “the latest inadvertent disclosure risk”, and asks firms to review their AI policies and training against it.
ACCA and the CCAB
ACCA applies the same principles. With CISI it published an AI Monitor report on ethical threats in October 2025, calling for AI literacy, governance frameworks and aligned policies. In June 2026 the CCAB, whose members include ICAEW and ACCA, published a draft statement to the profession on the use of AI for consultation. It builds on the IESBA Code. Check its status before you rely on it.
The Code of Ethics principles AI touches
- Confidentiality. R114.1 asks you to respect the confidentiality of client information and be alert to inadvertent disclosure. Pasting client records into a public AI tool is exactly that.
- Professional competence and due care. Understand what your tools can and can't do, check their output, and keep your CPD up to date.
- Objectivity. Don't let automation bias replace your judgement. Ask suppliers how their data and models handle bias.
- Integrity. Be open with clients about how you use AI, and don't steer a tool to justify an answer you'd already decided on.
- Professional behaviour. Presenting an AI error as fact can discredit the profession. Know the law around AI, including data protection.
Anti-money laundering and AI
The Money Laundering Regulations 2017 place duties on the firm for risk assessment, customer due diligence, record keeping and reporting suspicion through your MLRO. AI can collect documents, run checks in your existing ID software and flag exceptions, while a person makes the compliance decisions. Keep all suspicious activity report material out of AI tools because of the tipping-off offence.
UK GDPR
- Your practice is the controller and the AI provider is usually your processor, so you need a contract.
- Assess the risk before you start. AI on client data will often call for a data protection impact assessment.
- Know where data is stored and whether it leaves the UK, and mention AI in your privacy notice.
- Report personal data breaches to the ICO within 72 hours where required. The ICO's guidance on AI and data protection is under review after the Data (Use and Access) Act 2025.
Before you put client data into any AI tool, check where your practice stands on data, systems and policy. Our AI readiness checklist scores it in a few minutes and shows the gaps to close first.
AI policy for accountants: a free template
ICAEW asks firms to set clear AI policies, communicate them and train staff. This one is short on purpose. Replace the text in square brackets. The law firm version goes further on supervision and review if you want more detail.
- Owner: [Partner or director name]
- Effective from: [date]
- Next review: [date, no more than six months away]
1. Purpose and scope
- 1.1This policy sets out how everyone at [Practice name] may use artificial intelligence (AI) tools for practice work, including generative AI such as chat assistants, drafting tools and bookkeeping automation.
- 1.2It applies to partners, staff, subcontractors and anyone else working on our clients' affairs, on any device.
- 1.3It applies alongside the [ICAEW / ACCA] Code of Ethics, our data protection and information security policies and our anti-money laundering procedures. Where they conflict, the stricter rule applies.
2. Approved tools
- 2.1Only use AI tools on the approved list, kept at [location] by [role]. The list says what each tool may be used for and what data may go into it.
- 2.2A tool is only approved once [role] has confirmed that the provider does not train its models on our data, that data is held securely and deleted on our instruction, that we have a data processing agreement, and where data is stored.
- 2.3Do not use personal or free AI accounts for client work.
3. Client data and confidentiality
- 3.1Treat anything you put into an AI tool as disclosed to its provider. Entering client information into an unapproved tool is a confidentiality breach, even if you meant no harm.
- 3.2Client information may only go into approved tools that the list marks as cleared for client data.
- 3.3Never put information about a suspicious activity report, or an internal report to our MLRO, into any AI tool.
- 3.4Remove names and identifying details first wherever the task allows.
4. Review and sign-off
- 4.1AI output is a first draft. Review it with professional scepticism, as you would work prepared by a less experienced colleague.
- 4.2Check figures against the source records and check tax and legal points against primary sources before anything is relied on or sent.
- 4.3A qualified person signs off accounts, returns and advice, the same as without AI. Filing stays in our filing software and a person submits it.
- 4.4Record on the client file when AI was used for substantive work and who reviewed it.
5. Telling clients
- 5.1Our engagement letter tells clients that we may use AI tools, that a qualified person reviews the work and how we protect their data.
- 5.2If a client asks how AI was used, we answer honestly. If a client asks us not to use AI on their work, we follow that instruction.
6. Anti-money laundering
- 6.1AI may gather and check customer due diligence information in our approved ID and AML tools. Risk assessments and decisions stay with a person, and suspicion is always reported to our MLRO.
7. Training
- 7.1Everyone completes [the practice's AI training] and confirms they have read this policy before they get access to an approved tool. [Role] records who has been trained and when.
8. Incidents
- 8.1Tell [name] straight away if client information has gone into an unapproved tool or an AI error has reached a client or HMRC.
- 8.2[Name] records the incident, assesses the harm and decides whether to tell the client, report a personal data breach to the ICO (within 72 hours where required), notify our insurers, and report to our professional body where the Code of Ethics requires it.
9. Review
- 9.1[Name] reviews this policy and the approved list at least every [six] months, and after any incident or new guidance from [ICAEW / ACCA] or the ICO.
- 9.2Acknowledgement: I have read and understood this policy. [Name, signature, date].
How we set AI up within it
We configure AI to handle drafting and client chasing while a qualified accountant on your team signs off. Client data stays in your Microsoft or Google tenant, or our UK-hosted environment, using business AI tiers that do not train on your data. Figures link back to their sources, your MLRO makes the AML decisions, and filing remains in your MTD software.
