Can ChatGPT Pass the CPA Exam? Here’s What Happened

The CPA exam is one of the most challenging professional tests, requiring deep accounting knowledge, critical thinking, and application of complex financial principles. With AI models like ChatGPT becoming increasingly sophisticated, many wonder if they can handle such a rigorous assessment.

To find out, we tested whether ChatGPT could pass the CPA exam under realistic conditions.

The Exam’s Four Parts

The CPA exam is divided into four sections, each assessing different aspects of accounting and finance. We evaluated ChatGPT’s ability to answer questions from each section, covering topics from financial reporting to regulatory compliance.

Auditing and Attestation

This section tests auditing principles, financial statement evaluation, and internal controls. It requires knowledge of Generally Accepted Auditing Standards (GAAS) and Public Company Accounting Oversight Board (PCAOB) regulations. Topics include risk assessment, fraud detection, and evidence collection.

Candidates must determine compliance with accounting standards, such as revenue recognition under ASC 606, which governs how businesses account for contracts with customers. Another common task involves evaluating internal controls using the COSO framework, a widely used model for assessing risk management and fraud prevention.

Business Environment and Concepts

This section covers corporate governance, economic principles, financial management, and information technology. Candidates must understand how macroeconomic factors like interest rates and inflation influence financial decision-making.

A key topic is cost-volume-profit (CVP) analysis, which helps businesses determine the sales needed to cover costs and generate profit. Candidates may calculate a company’s break-even point in units or dollars, requiring knowledge of fixed and variable costs. Capital budgeting techniques, such as net present value (NPV) and internal rate of return (IRR), are also tested, as they are essential for evaluating long-term investments.

Financial Accounting and Reporting

This section assesses financial statement preparation under Generally Accepted Accounting Principles (GAAP) and International Financial Reporting Standards (IFRS). Topics include asset valuation, liabilities, equity transactions, and income recognition.

A frequently tested area is lease accounting under ASC 842, which requires companies to classify leases as operating or finance leases and recognize right-of-use assets and liabilities on the balance sheet. Deferred tax accounting under ASC 740 is another focus, requiring candidates to compute deferred tax assets and liabilities based on temporary differences between financial and tax reporting.

Regulation

This section examines tax laws, business law, and professional ethics. Candidates must be familiar with the Internal Revenue Code and various tax provisions, including individual, corporate, and partnership taxation.

Capital gains taxation is a key topic, with short-term gains (on assets held for one year or less) taxed at ordinary income rates, while long-term gains receive preferential rates ranging from 0% to 20%, depending on taxable income. Payroll tax compliance is also tested, requiring knowledge of Social Security and Medicare tax rates (6.2% and 1.45%, respectively, for employers and employees). Ethical considerations focus on the AICPA Code of Professional Conduct, which outlines accountants’ responsibilities regarding integrity, objectivity, and confidentiality.

Passing Score Criteria

The CPA exam uses a scaled scoring system, with each section graded from 0 to 99. A minimum score of 75 is required to pass. The exam employs a weighted scoring model, meaning correctly answering more difficult questions contributes more to the final score than easier ones.

Multiple-choice questions (MCQs) make up a significant portion of the exam and are graded using adaptive testing. As candidates progress, the difficulty of subsequent questions adjusts based on prior responses. Task-based simulations (TBS), which require applying knowledge to real-world scenarios, are also heavily weighted and often test more complex concepts. Written communication tasks, found in certain sections, assess a candidate’s ability to clearly and concisely convey financial information.

Scoring incorporates professional judgment. In TBS questions, partial credit may be awarded for correct calculations or proper application of accounting principles, even if the final answer is incorrect. This ensures that candidates demonstrating strong analytical skills can still earn points despite minor computational errors.

AI Considerations

Artificial intelligence has advanced in processing financial data, but the CPA exam presents unique challenges. Unlike structured financial modeling or tax preparation software, which operate within predefined parameters, AI must interpret nuanced questions that require layered reasoning and professional judgment.

One challenge is understanding the context behind financial decisions. Many CPA exam questions require assessing the best course of action based on competing priorities. For example, when evaluating goodwill impairment under ASC 350, human candidates consider qualitative factors, such as industry downturns or declining cash flow projections, before performing the quantitative impairment test. AI, however, relies solely on the information provided and struggles to infer broader economic implications without explicit data.

Legal interpretation is another difficulty. Tax law is filled with exceptions, phase-outs, and evolving case law that require careful reading beyond applying statutes. The Tax Cuts and Jobs Act introduced limitations on net operating loss (NOL) carryforwards, which were later modified by the CARES Act and adjusted again in subsequent legislation. A CPA must track these changes and apply the correct rules based on the taxpayer’s filing year. AI, unless continuously updated with real-time regulatory changes, risks misapplying outdated provisions.

Ethical dilemmas also pose a challenge. The CPA exam often presents scenarios where candidates must balance professional responsibilities with client interests, applying guidance such as the AICPA’s Conceptual Framework for Independence. A human accountant may recognize when a conflict of interest exists even if no explicit rule is broken, whereas AI struggles with the subjective nature of ethical reasoning. This is particularly relevant in auditor independence, where maintaining public trust is as important as following the law.