While AI offers recognised benefits, including improved efficiency and speed, as well as new analytical capability, these need to be considered alongside ethical risks and the responsibilities that continue to define professional accounting practice.
A draft statement to the profession from Ireland’s Consultative Committee of Accountancy Bodies warns that AI tools may ‘create, exacerbate or obscure’ ethical risk, with overreliance on AI potentially weakening critical thinking, professional scepticism and independent judgment. Recent high-profile reports of firms issuing apologies for their use of AI, combined with research highlighting ‘healthy scepticism’ among professionals about AI adoption, suggests that this is an important moment to return to first principles.
Ethics
Recent guidance from the International Ethics Standards Board for Accountants (IESBA) considers not only generative and agentic AI, but also other cutting-edge technologies, and reinforces an important point: emerging technologies do not reduce the need for professional judgment, they increase it.
AI has the potential to amplify threats
Ethical threats for accountants are, of course, nothing new. They include financial or other interests inappropriately influencing judgment or behaviour, failure to appropriately evaluate the results of a previous judgment or activity, the promotion of a client’s or organisation’s position compromising objectivity, long or close relationships leading to excessive sympathy for client/employer interests, and actual or perceived pressure deterring an accountant from acting objectively.
AI has the potential to amplify these threats. For example, pressure to work quickly can discourage proper checking, while familiarity with a trusted system may lead to its outputs being accepted without sufficient challenge.
Evaluating and addressing ethical threats requires an ‘inquiring mind’
These ethical threats are effectively countered by IESBA’s international code of ethics, which professional accountants must abide by. Identifying, evaluating and addressing the ethical threats requires the possession of an ‘inquiring mind’, the exercise of professional judgment and the application of the ‘reasonable and informed third-party’ test (ie the conclusion that would be reached by an appropriately informed and impartial person after considering the relevant facts and circumstances).
The code sets out five fundamental principles (integrity, objectivity, professional competence and due care, confidentiality, and professional behaviour) that give accountants a clear way to tackle the threats, as follows.
Integrity
The integrity principle requires honesty about both the use of AI and its limitations. AI-generated content should not be presented as reliable or authoritative unless it has been properly checked. In a profession built on trust, transparency about how information is produced is fundamental. Recent withdrawals and corrections of AI-assisted reports by Deloitte (in Australia), KPMG (globally), EY (in Canada) and PwC (in its Middle East business) illustrate the risk. In each case, AI-generated references, citations or factual claims appear to have passed through review processes without being adequately challenged. AI outputs should be treated as inputs to professional judgment rather than definitive answers in their own right.
Objectivity
Professional accountants cannot assume that technology is objective simply because it is automated. AI systems may reproduce existing biases, assumptions and omissions in their data or design. In addition, accountants must guard against automation bias (ie the tendency to place undue confidence in a machine-generated answer because it appears precise or impartial). Accountants must critically evaluate AI outputs, challenge unexpected results and consider how prompts and data have shaped the output while resisting the use of AI to confirm a conclusion already preferred.
The misuse of AI can undermine trust that has taken decades to build
More information
Read ACCA’s new professional insights report, Navigating ethics for trust
Join a session discussing the report’s findings at ACCA’s flagship Accounting for the Future conference. Register to watch live on 24-26 November or on demand
Professional competence and due care
The professional competence and due care principle requires accountants to understand enough about an AI system’s capabilities and limitations to judge whether it is appropriate for the task at hand. Checking an AI-generated report involves more than proofreading it. Figures, assumptions, sources and conclusions may all require independent verification. As systems, data and models change, their continued reliability may also need to be reassessed. Professionals should not surrender judgment to technology; they should instead use technology to enhance judgment.
Confidentiality
Clients, employers and stakeholders share information with accountants on the understanding that it will be handled responsibly. The use of AI does not diminish this obligation. Entering a client spreadsheet into a public or unapproved AI tool, for example, could disclose commercially sensitive information. Before using an AI system, accountants should understand how the data will be stored, processed and protected, and comply with organisational policies and data protection requirements.
Professional behaviour
An action can be legal or technically possible and still fall short of the conduct expected of a professional accountant. Accountants must consider how their use of AI may affect clients, other stakeholders and confidence in the profession. The misuse of AI, careless reliance on unverified outputs or inappropriate handling of confidential information can undermine trust that has taken decades to build.
Human agency
The five principles show that responsible AI use is more than a technical issue. It is a matter of professional judgment, governance and accountability.
Responsibility cannot be delegated to technology
Perhaps the most important point of the IESBA code is that responsibility cannot be delegated to technology. AI may perform analysis, draft a report or generate a recommendation, but accountability remains with the accountant.
This is an exciting period of change for the profession. AI offers considerable opportunities to improve efficiency, insight and decision-making. However, our ethical responsibilities remain unchanged. The real challenge lies in continuing to question outputs, protect confidential information and exercise the judgment and ethical leadership expected of trusted professionals.
AI’s challenges for accountants
- A lack of transparency in how AI reaches a conclusion (‘black box’ behaviour) can make it difficult to identify errors, bias or inappropriate assumptions, and to explain output to clients or other stakeholders.
- AI models may contain algorithmic bias if trained on skewed, incomplete or erroneous data, creating risk around its use.
- AI systems may present factually incorrect, fabricated or unsupported information (‘hallucinations’) with considerable confidence, making unreliable outputs appear credible.
- Entering confidential or commercially sensitive information into an AI system may result in unauthorised retention, reuse or disclosure.
- It may not always be clear who reviewed output to which AI contributed, and who is responsible for the final judgment.
- Research on algorithm appreciation indicates an emerging preference for AI-generated recommendations over human judgment.
