Nobody asks an accounting firm which tax software it uses.
That observation led Adam Shay, CPA, to pose a timely question on LinkedIn: If clients have accepted technology as part of accounting work for decades, why should artificial intelligence require a different level of disclosure? After all, clients generally care about the result. They want an accurate tax return, clean books, reliable financial information, and sound advice. How accounting professionals produce that result has traditionally been their responsibility — not something clients are expected to investigate.
AI, however, is testing that assumption.
Shay points to a Karbon survey of 350 small-business owners and leaders who personally manage relationships with external accounting firms or CPAs. According to the research, 89% want some transparency about how and when AI is used in their work, and 57% want full transparency. Meanwhile, only 21% of accounting firms have a documented AI policy or strategy. Accounting leaders can debate how much disclosure is appropriate. What they can no longer afford to do is avoid the question.
AI introduces a different kind of risk
Shay’s comparison between AI and tax software is useful, but there is an important difference between the technologies. Traditional accounting systems are generally deterministic and designed to automate defined tasks. When given the same data and instructions, they are expected to produce the same result. Errors can still occur because of faulty data, incorrect configurations, software defects, or user mistakes, but the system itself is intended to operate predictably.
AI is generally more nondeterministic. Its output can vary, even when a user provides similar input and instructions. Gen AI can also generate information that sounds authoritative but is incomplete, misleading, or wrong. Highly publicized examples of AI hallucinations have made that risk visible to the public in a way that the risks associated with most other accounting technologies are not.
That visibility helps explain why clients view AI differently. They are not necessarily asking for disclosure because they want to scrutinize an accounting firm’s technology stack. They want enough information to assess the possibility that AI could contribute to inaccurate work or bad advice.
This does not mean accounting firms must identify every AI-enabled feature embedded in every platform they use. As AI becomes integrated into tax, audit, financial reporting, practice management, and productivity software, such an inventory may become both impractical and unhelpful to clients. But clients still need enough information to understand how the firm manages the risks created by those tools.
Clients want efficiency, but they still expect human judgment
Shay also highlights an important distinction in the Karbon findings: Clients appear relatively comfortable allowing AI to perform certain tasks, but they are less willing to place judgment and trust in its hands. According to Shay’s summary of the survey, 61% of respondents want the trusted adviser role handled entirely by a human, while 58% say the same about crisis management. Clients may accept AI identifying anomalies, processing information, or supporting routine work. They are less comfortable with AI interpreting a complicated situation, delivering difficult news, or guiding them through a consequential decision.
That should not be interpreted as a rejection of AI. It is a reminder of what clients believe accounting professionals are there to provide. Clients do not turn to accountants only for calculations. They rely on them for context, professional judgment, reassurance, accountability, and advice grounded in an understanding of their goals and circumstances. While AI may help professionals deliver those services more efficiently, it cannot assume professional responsibility for the quality of services. When AI contributes to an analysis or recommendation, a qualified professional must still evaluate the output, consider the relevant facts, and stand behind the conclusion or recommendation. “The AI generated it” will never be an adequate explanation when a client receives incorrect advice.
Quality management provides the better answer
The most productive response to clients’ concerns is not simply, “Yes, we use AI.” Nor is it a lengthy list of technologies. The stronger response is: “Yes, we use AI, and we have a system for managing its risks.”
A good accounting firm has a strong system of service quality management. Policies, review procedures, documentation requirements, consultation protocols, and clear lines of accountability help ensure that work is performed consistently and meets the profession’s standards. Quality management is most commonly associated with audit and assurance services. Tax practices also tend to have established review processes, particularly when preparing returns, conducting research, and advising clients about complex matters. However, the underlying principles should not stop there.
Client accounting services, outsourced finance functions, consulting, and other advisory services are increasingly relying on AI-enabled technologies. Those services can affect strategic decisions, cash flow, financing, compliance, financial reporting, and the future direction of a client’s business. If AI introduces risk into that work, quality management must extend to it.
The same principle applies beyond public accounting. Corporate finance teams, nonprofit organizations, government entities, and other accounting professionals may use AI to analyze information and support decisions. Their stakeholders may include executives, boards, investors, regulators, donors, or taxpayers rather than external clients, but those stakeholders still need confidence in the integrity of the work. Quality management should follow the risk, regardless of the service line or organizational setting.
A policy is only the beginning
An AI policy is an essential component of quality management, but having a document is not enough. Organizations also need procedures that translate the policy into daily decisions.
Leaders should be able to answer several fundamental questions:
- Where and how is AI being used?
- What client, financial, or confidential information may be entered into an AI system?
- Which tools are approved and which are prohibited?
- Which outputs require human verification?
- How should employees document their use of AI?
- When is consultation with a subject matter expert required?
- Who reviews and approves the final work?
- How are errors or unexpected outputs reported and corrected?
- What information will be shared with clients and other stakeholders?
These questions become especially important as AI capabilities are added to software employees already use. People may not always recognize when an established workflow begins relying on AI or understand how that changes the risk. Training, monitoring, and ongoing communication are therefore as important as the written policy itself. Organizations cannot provide meaningful transparency if their leaders do not know how AI is being used.
Transparency should reinforce confidence
AI disclosure does not have to sound like a warning. When handled well, it can demonstrate that a firm understands the technology, anticipates its risks, and has established appropriate safeguards.
A client-facing explanation might say:
We use approved AI-enabled technologies to support portions of our work and improve the efficiency of our services. We also follow the accounting profession’s quality management and ethical standards, protect confidential information, review relevant AI-generated outputs, and apply professional judgment before delivering our work. Qualified professionals remain accountable for the services and advice we provide.
The precise language will depend on the services performed, the organization’s actual practices, applicable professional requirements, and advice from legal, risk, and insurance professionals. Whatever language an organization chooses, it must reflect reality. Transparency creates trust only when the controls described to clients are consistently followed.
The real disclosure is accountability
Clients are not asking for a guided tour of an accounting firm’s technology stack. Nor are organizational stakeholders asking for a review of the finance team’s technology stack. In both cases, they want to know whether the professionals handling their work understand the risks, protect confidential information, verify technology-generated results, and remain accountable.That is the more useful way to frame the AI disclosure debate.
Accounting professionals may reasonably reach different conclusions about when and how to disclose AI use. They should base those decisions on the nature of the service, the significance of the AI-assisted work, the potential consequences of an error, and the client’s or stakeholders’ reasonable expectations. But failing to respond to the “Did you use AI?” question is no longer a responsible position. AI is already influencing accounting work. Clients know it, employees are experimenting with it, and software providers are embedding it into familiar platforms. The organizations that preserve trust will not necessarily be those that disclose the most or use AI the least. They will be those that can explain, clearly and confidently, how human professionals remain in control and ensure the quality of work.
| Put Responsible AI Into Practice Using AI effectively requires more than choosing the right technology. Explore the ethical standards, quality controls, and strategic thinking accounting professionals need to manage AI responsibly. Members use promo code [ENTER YOUR PERSONALIZED PROMO CODE FOR YOUR MEMBERS] and save 20% on all course registrations with the Center for Accounting Transformation. Dec. 8: How Accountants’ Ethics Are Impacted by the A-B-C-Ds of the Digital Age Examine the ethical implications of artificial intelligence, automation, blockchain, cybersecurity, and data governance — and learn how accountants can help mitigate emerging risks. This two-hour course offers 2 CPE credits. Dec. 9: Using AI Responsibly: Ethics, Accuracy, Privacy, and Professional Judgment Explore the safeguards accounting professionals need to protect sensitive information, evaluate AI-generated content, and preserve human accountability. Dec. 15: How to Turn AI and Automation Into Strategic Decisions Move beyond AI FOMO. Learn how to align technology investments with organizational strategy, evaluate business value, and balance innovation with risk. This one-hour course offers 1 CPE credit. Explore more learning opportunities from the Center for Accounting Transformation. HSCPA code is for 20% off, so I am guessing any org that has an MOU will have the same discount, but I am not quite sure who on our team to ask… |
Sign in to get access to this free resource, and all of our whitepapers and reports.
Download this content today!
Register Now Already registered? Click here to Log In
