The Consultative Committee of Accountancy Bodies (CCAB) has unveiled a new 'AI hub' for accountants, including a draft statement and case studies on the ethical use of Artificial Intelligence (AI). The CCAB emphasises that an accountant remains responsible for work produced by AI. The committee provides some practical tips on using AI tools and systems.
CCAB's draft statement on ethical Artificial Intelligence (AI) use, together with a series of case studies, shows how accountants can fulfil their ethical obligations when using the new technology in their professional activities.
- The core principle is that when using AI, accountants remain fully responsible and accountable for their work.
The draft statement and case studies are themselves an example of the use of AI. The Frequently Asked Questions (FAQs) sections of the documents explain how and why this was done.
- Feedback on the statement and the case studies is invited through the use of short surveys included in the documents.
Here is a summary of the key points contained in the statement.
Ethical risks
The statement identifies seven common risks in using AI:
- Algorithmic bias.
- The data that AI models are trained on could be biased, incomplete or factually wrong.
- Automation bias.
- People tend to favour automated outputs over human reasoning.
- Data privacy and confidentiality.
- Use of public AI models could lead to unauthorised disclosure, loss of legal privilege or unauthorised use for training of AI models.
- Failure to adopt an inquiring mindset, erosion of professional judgement and a loss of professional scepticism.
- Hallucinations.
- AI can provide plausible, but factually incorrect responses.
- Opacity.
- The difficulty in tracing data sources or understanding the logic behind AI-generated conclusions.
- Unclear accountability for outputs produced by AI and loss of trust in the profession.
The CCAB stresses the need for accountants to exercise rigorous professional judgment when using AI tools and systems and to apply an 'Inquiring Mindset' when relying on their output. Responsibility for the work produced cannot be abdicated.
Applying fundamental principles
The statement explains how accountants can comply with the five ethical fundamental principles when using AI.
Integrity
- Accountants should:
- Be transparent with clients about the use of AI tools on client engagements.
- Ensure that AI-generated reports do not contain materially false or misleading information.
- Avoid 'recklessly' associating themselves with AI work without adequate validation.
- AI systems need to allow for contestability and redress, since accountants must take steps to disassociate themselves from false, misleading or recklessly produced information when they become aware that this is the case.
Objectivity
- Scrupulous checking of AI-quoted sources is required to mitigate automation bias.
- Awareness is needed of the potential impacts of how prompts entered into an AI tool are framed. Accountants need to ensure they do not enter prompts that are designed to produce biased or unfair outcomes.
Professional competence and due care
- Accountants need to maintain a continuing awareness of technology-related developments.
- Skills required of accountants include:
- Being able to use AI tools and systems effectively and ethically.
- Ensuring that clients receive services based on current technical standards.
- Understanding AI's inherent limitations.
- Being able to explain to the client any material limitations on the use of an AI tool or the outputs it produces.
Confidentiality
- Safeguards need to be in place for data input into, processed by or retained within AI systems.
- Client or any sensitive data should never be input into an AI tool without specific consent. Particularly sensitive areas include:
- Data may be kept within an AI tool for training purposes.
- Data might end up entering the public domain.
Professional behaviour
- Use of AI must comply with relevant laws, including those governing data protection and intellectual property.
- Any AI-generated outputs need to be fully checked and validated before being:
- Submitted to a Court.
- Placed in the public domain.
- Published in circumstances where others may rely on the contents.
Threats and safeguards
The CCAB also provides advice on how to handle threats in complying with the fundamental principles arising from AI use.
- Threats include self-interest, self-review, advocacy, familiarity and intimidation.
- The assumption of management responsibility could also be a threat when acting for audit clients.
Potential safeguards include having robust internal controls, third-party assessments of AI functionality and comprehensive staff training on AI ethics and prompt engineering.
- If safeguards cannot be used to eliminate threats or reduce them to an acceptable level, accountants are expected to decline or discontinue the professional activity.
Responsible use
The CCAB expects accountants to establish clear ethical and business guidelines when developing and implementing processes for using AI systems. AI systems are to be used to enhance, not replace, human judgment.
- Aspects that should be considered include transparency, accountability, fairness and maintaining human oversight to correct or rectify AI decisions.
Public interest
Because accountants are ethically required to act in the public interest, decisions about AI use cannot be based solely on efficiency or commercial considerations.
- Vendors should be challenged to provide 'model cards' that specify the risks and biases in their AI systems.
- It should be ensured that AI use does not propagate disinformation and fraud.
The EU Artificial Intelligence Act
Professional accountants who practise in or do business with the EU will need to be familiar with the EU Artificial Intelligence Act.
Use of AI Agents
AI agents are capable of acting with little human intervention, which raises heightened ethical concerns.
- Accountants must ensure that any AI agents operate within a 'constrained autonomy' framework defined by the accountant.
- Clear human responsibility for the use and effective oversight of an AI agent should be retained.
- Tasks can be delegated, but the accountant remains professionally responsible.
- Appropriate safeguards, including 'kill switches', need to be in place and rigorously monitored so that an accountant can intervene and prevent an ethical breach.
- If representing the accountant at a meeting or in a public setting, care is needed to ensure the AI agent does not breach principles of professional behaviour or act in a manner likely to discredit the profession.
Practical guidance
The statement concludes with some practical tips for using AI tools and systems ethically, including:
- Disclosing the use of AI.
- Use secure environments.
- Only using enterprise-grade tools that guarantee data privacy and do not use data for model training.
- Use ethical prompts.
- Always fact-check and verify the accuracy of AI-generated data.
- Maintain a 'human in the loop'.
- Ensure there is a process that allows individuals to challenge AI-driven decisions.
- Stay informed about emerging AI risks and best practices through professional body resources. Invest in training to understand how AI works, including the art of 'ethical prompting'.
Case studies
The draft case studies are fictional scenarios intended to highlight potential ethical dilemmas. They cover:
- Professional accountant in business - AI-generated strategy and planning.
- Professional accountant in practice - AI in audit evidence collection.
- Professional accountant in a large firm - synthetic client data for LLM training.
- Professional accountant in the public sector - AI in public expenditure.
- Professional accountant as NED - AI in financial reporting.
- Professional accountant in Academia - AI in accounting education.
Useful guides on this topic
Professional Conduct in Relation to Taxation
The Professional Conduct in Relation to Taxation (PCRT) sets out the ethical and professional standards expected of members of the seven authoring professional bodies when advising on UK tax matters.
New guidance published on use of AI in tax
Topical guidance on using Artificial Intelligence (AI) in tax work has been issued by the seven professional bodies responsible for producing the document 'Professional Conduct in Relation to Taxation' (PCRT). The new guidance is designed to be read in conjunction with existing PCRT guidance and discusses scenarios and possible safeguards allowing AI tools to be used ethically.
Accountants' misuse of AI in FTT case
In Gary Elden v HMRC [2026] TC09742, the First Tier Tribunal (FTT) found that significant procedural failings, exacerbated by the unverified use of Artificial Intelligence (AI) by Mr Elden's accountants, posed serious risks to the fairness of proceedings. Nevertheless, it refused HMRC's strike-out application.
External link