AI can simplify complexity in litigation support

Artificial intelligence can enhance litigation support by helping accounting experts simplify complex accounting concepts and present compelling evidence. Here’s how to navigate the technology’s limitations, legal implications and best practices for using AI to advance justice.

Financial fraud cases are rife with complex terms and reams of documents and data. To the accountants and other financial experts who spend months or years working through the data and following money trails to the perpetrators, these circuitous details make sense. But consider what an alleged construction fraud case involving multiple overlapping contracts — agreements between a contractor and client and a separate loan agreement with a bank — might look to a jury without the expertise to parse it. An accountant testifying as an expert witness might as well describe the evidence in an obscure language. As fraud cases become increasingly data-driven and complex, the role of financial analysis and litigation support experts must evolve. In the courtroom, accountants are no longer just number crunchers. They must be storytellers, translators and key players in the pursuit of justice. 

When legal disputes arise, it’s up to accounting experts to dissect the numbers, untangle conflicting interpretations and present a clear picture of what happened. In a construction fraud case, anti-fraud professionals called to testify in court must identify who agreed to what by collecting a paper trail that often spans hundreds of pages and entails contracts, loan agreements, draw requests, inspector reports and other documents. Their work also involves reconstructing a detailed, month-by-month narrative of the project’s progress measured against financial records and contract terms. Presenting dense financial concepts in an accessible format for a jury is challenging. In the courtroom, clear communication is just as critical as technical accuracy. Considering the heavy task they have, many experts have turned to artificial intelligence (AI) tools for help in transforming dense data into digestible, visually compelling narratives that can withstand legal scrutiny. 

Experts in the courtroom must navigate competing narratives — one grounded in accounting standards and contractual expectations, the other shaped by cash flow timing, cost estimates and project delays. Traversing between these perspectives requires a deep understanding of accrual and cash accounting and the ability to clearly explain technical nuances, including variance analysis, Generally Accepted Accounting Principles (GAAP) compliance and Accounting Standards Codification 606. Additionally, legal disputes often involve vast amounts of private and public information.

AI is redefining how financial analysis and litigation support experts approach their craft in high-stakes settings, including civil and criminal litigation. Going beyond streamlining workflows, AI can turn esoteric data into stories that inform and persuade — if used properly. For instance, in a fraud case, AI can help accountants analyze data more efficiently by sifting through massive datasets and years of transactions and produce a report that highlights suspicious patterns, such as inflated or opportunistically timed revenues. AI tools can also condense a 100-page financial report into a 500-word summary or series of bullet points. They can easily replace overly technical terms, such as “deferred tax liability,” with simplified wording, such as “future tax obligations.” AI can also add bar graphs or pie charts to financial datasets to make them more visually appealing and accessible, such as charts comparing cash flow projections and discount rates in business valuation disputes. In a recent case, a law firm used AI to analyze and organize 63 depositions given by a single expert along with more than 7,500 pages of text to determine the consistency of the expert’s testimony over time. 

 
With proper guidelines and an accurate understanding of how to use it, AI can be an indispensable ally for accounting professionals tasked with creating reports, summaries and briefs for litigation (see Table 1 below). We’ll explain how experts may use AI effectively to streamline communication and transform complicated financial data into usable, persuasive evidence. 

Table 1: AI support for expert witness communication investigating a multi-period construction contract

AI foundations

Large language models (LLM) are AI systems trained to understand and generate human language. LLMs can write essays, generate computer code and solve problems. They’re built upon data collected from training, books, websites and articles. In 2022, OpenAI released ChatGPT, a generative pretrained transformer (GPT) LLM type of AI capable of generating human-like text based on user input. Other LLM-based AI products include Google Gemini, Claude and CoPilot. 

The more people use AI, the more information it’s able to collect for its repository. Frequent use of AI provides feedback that allows it to fine-tune its responses. Although many AI products have free versions, subscription-based models provide more up-to-date learning with access to domain-specific resources and allow users to customize for specific uses. A typical paid AI subscription offers higher or unlimited usage, advanced features, technical support, and enhanced privacy and security.

Since the launch of ChatGPT, the use of AI as a professional tool has constantly increased. A survey conducted by McKinsey’s Global Survey Panel shows that 65% of businesses used generative AI in 2024, up from 33% in 2023.

Getting started with AI in accounting

Many remain skeptical about using AI for high-stakes work, such as in civil and criminal litigation. To reduce skepticism, users must familiarize with how AI works. As tools for data analysis, discussion generation and technical reporting become mainstream, AI literacy is integral to professional competency.

To get started with AI, you must craft precise instructions, in a process known as prompt engineering. At first, many people anticipated that prompt engineering would be exclusive to trained experts. However, AI technology has rapidly evolved into a much more accessible tool and has become better at recognizing conversational language.

An effective prompt should define the task, context, audience and response format. A prompt can be further refined to specify tone, style, length, quantity or even the language of a desired response. Users can provide examples or templates of outputs and include instructions in the prompt to avoid certain verbiage in the output (e.g., avoid color schemes in graphs that are difficult for individuals with color blindness to differentiate). In addition to responding to a prompt, AI will often ask clarifying questions to ensure that the outcome is what the user is looking for. 

AI prompt engineering can generate courtroom-appropriate content. The Five “S” Model, developed by AI literacy organization AI for Education, provides guidance on how to structure effective prompt engineering.

• Set the Scene. Describe the accounting dispute or audit context (e.g., misstated revenue, GAAP accounting treatment, forensic reconstruction, asset tracing).
• Be Specific. Define the disputed economic issues within the accounting context (e.g., GAAP accounting treatment or tax code compliance, materiality thresholds, asset concealment, inventory valuation).
• Simplify Your Language. Ask AI to explain complex topics such as revenue recognition, book/tax differences, deferred tax assets or consolidation adjustments in plain terms.
• Structure the Output. Request response formats suited for exhibits (e.g., side-by-side comparisons, step-by-step reconciliations, flowcharts, color-coded graphs).
• Share Feedback. Refine outputs provided by AI with clarifying language to address misinterpretations and to align the explanation with legal briefs, direct testimony and the cross-examination strategy.

Clarity is key when writing prompts. For instance, instead of saying, “Tell me about taxes,” try inputting, “Summarize the tax implications of accelerated depreciation for a small-business owner in plain language understandable for a beginner.” Additional refinements may include adding context, such as, “Use examples from retail businesses;” constraints, such as, “Limit to three key points;” or specifics, such as, “Provide slides for a PowerPoint presentation.”

To use AI to analyze complex data in a spreadsheet, such as a trial balance with anomalies, consider beginning a prompt with, “Identify unusual patterns in this spreadsheet,” and then iterate to, “Highlight transactions exceeding $10,000 with explanations suitable for a jury.” AI users should keep a log of prompts and responses to track what works; this helps build intuition for the phrases that will avoid generating responses full of jargon. Patience is essential; initial outputs will most likely need tweaking, but with practice, you’ll craft prompts that consistently deliver tailored, digestible content in narrative form or as a chart-ready summary.

Another way to use AI is to improve writing. If you have a report already written, you can use AI to check for redundancy or help change the tone of the writing. For example, prompt AI with, “What recommendations would you make to update the tone of this legal brief for a judge and jury?” Or simply instruct AI to “update the tone of this brief for a legal setting with a persuasive message for a jury.”

An accountant’s work products can be expansive. Accountants can save time by using AI’s recommended simplified terminology (see Table 2 below) to create accessible summaries, briefs and data visualizations in their testimony prep. 

Table 2: ChatGPT response to simplify financial concepts

AI can generate alternative phrases for any brief, supported by graphics, to create simple, persuasive presentations. Other AI-based tools, such as Grammerly or ProWritingAid, integrate word-processing programs and provide suggestions for sentence structure, such as repetitive wording that may be distracting or active versus passive voice. Using multiple AI products to get started and then fine-tuning with Grammerly or ProWritingAid can dramatically speed up the writing process.

Likewise, AI can help with deposition and trial preparation. Upload a complete report and ask AI to provide sample questions that could be used for direct and cross-examination. Focus on a specific area of expertise and ask AI for details on how to prepare testimony.

For example, prompt AI to “provide a draft of a response to a question regarding Accounting Standards Codification 606 in a multiperiod construction of a commercial building from the perspective of a general contractor.” The summary AI provides can be a compelling preparation aid. You can then ask AI to “provide a series of questions that an accounting expert for the defense might be asked during cross examination in a fraud setting wherein accusations are based upon differences in cash and accrual expense accounting are the basis for the accusation.” 

AI limitations

AI isn’t infallible, and it has some known drawbacks, including some of the ways it responds to prompt engineering. One major drawback is the risk of “hallucinations,” or instances where AI generates plausible but inaccurate information, like misstating tax codes or citing completely erroneous source references. Hallucinations can mislead users who are analyzing complex data or explaining difficult accounting concepts. This is especially problematic in high-stakes settings such as litigation, where errors could influence how a proceeding is judged. 

Fake citations can negatively impact a court case and severely damage the reputation of experts. A Stanford professor submitted an expert legal declaration in November 2024 that contained fake legal citations, which caused opposing counsel to request that the court remove the declaration as evidence. And the consequences for filing information with a court using AI hallucinations can be severe. In August 2025, an Arizona judge removed an attorney from a case who presented information to the court with AI-hallucinated legal cases. The judge also required the attorney to notify the judges in the attorney’s other pending cases about the hallucinations in the Arizona case and referred the matter to the attorney’s state bar association for additional sanctions. It’s up to accounting experts to ensure the accuracy and proper use of AI in all work products.  

To mitigate these risks, use AI’s precision mode so that you’re only using verified information. Ask for accuracy by specifying the source you want to check and always cross-check AI results with primary records. Accountants should start with small, low-risk tasks, such as drafting a basic report, before moving on to critical analyses. Also, consider formulating independent judgments first and then comparing your judgments with those from AI. 

A best practice is to start by asking AI, “What is your knowledge cutoff date?” This will help determine whether current information or recent developments will impact the outcome you get from AI. A free version of AI may not have the most recent training data that a paid version would have, but even when using a free version of AI like ChatGPT, you can activate the web search feature to ensure it uses recent data to respond to your prompt and specify that it relies only on authorized sources when searching the web.

Using and creating custom GPTs

Once you’re comfortable with the basics of prompt engineering and familiar with the limitations of AI, you might consider finding or customizing a GPT for your unique needs. Users can “explore” ChatGPT for specialized GPTs. Options vary based upon use of the free or premium version. ChatGPT also allows users to create their own custom GPTs using a straightforward process that doesn’t require any programming knowledge.

To get started with a customized GPT, OpenAI’s “Create a GPT” tool walks the user step by step through the process of creating and defining the GPT’s behavior by inputting clear, detailed instructions. Users also have the option to upload specific source documents in multiple formats, such as accounting standards, industry-specific reports or sample financial exhibits, and ask the GPT to only rely upon those sources for a summary or analysis. Once set up, the customized GPT can be instructed to create responses that maintain a consistent tone tailored to the user’s requirements without additional contextual prompting.

The following compares generic AI (ChatGPT) with customized GPT (a mini app within ChatGPT):

ChatGPT (via ChatGPT app or “chat.openai.com”)

• Default, general purpose, standard AI.
• Gives solid responses; aims to be neutral and straightforward.

Custom GPT (mini app within ChatGPT)

• Custom-tuned AI with unique focus or specific personality (e.g., professional, highly detailed, precise, clarity-focused).
• Allows custom instructions (e.g., “Always answer like an accounting expert preparing responses for a court filing” or “Focus only on accounting topics from the following file”).
• Customized tools (e.g., focused web search, data analysis, image generation).
• Customized knowledge (e.g., responses draw from a specific PDF file or dataset uploaded by the user).
• Memory settings allow consistent reference to predetermined settings, sources and preferences.

Legal and ethical implications of using AI

The legal landscape for the use of AI is quickly evolving. Courts are considering various disputes regarding copyrighted material and the use of AI, and many jurisdictions, including the U.S. states of Utah and California, have passed disclosure laws regarding generative AI. The medical profession has identified various ethical issues concerning AI that appear relevant to accountants and expert witnesses, including accountability, empathy, explainability, privacy, consent, transparency and security.

Expert witnesses should ensure that their work complies with legal, ethical and professional requirements. Noting that AI can improve the quality and efficiency of legal services, the American Bar Association has provided ethical and practical guidance to its members. Likewise, many state bar associations have issued guidance, frameworks and rules regarding the use of AI in the legal process. Fraud experts are encouraged to consult these resources to ensure AI compliance in their jurisdictions. 

In a Journal of Accountancy article, risk consulting director J. Michael Reese suggests that accountants provide informed consent to clients regarding the use of AI to prevent disputes and misunderstandings. Specifically, informed consent allows clients to make decisions before work commences and enhances communication between the accountant and their client.

To use AI ethically, it’s important to prioritize data privacy, reduce bias and foster transparency. To prioritize privacy, when using AI, remember that data input into AI or conversations with AI may be stored and used for “training,” potentially exposing confidential data. Handling sensitive data, financial information or regulated information requires the use of an enterprise grade or locally hosted AI. The best practice when using any AI, whether free, purchased or customized, is to check the settings and turn off the training mode (if possible). Subscription models often offer more enhanced privacy and security. In free versions of OpenAI, conversations with ChatGPT held without an account are used for AI training by default. To opt out of conversations being used as training data, go to Settings and uncheck “Improve the model for everyone.”

Consider when using AI that the data it was trained on may be biased, misleading or incomplete, which can skew outputs from the system and perpetuate bias and injustice. In the world of litigation support, output must be factual. Failures or biases in the output can result in lost trials, miscarriages of justice and damaged reputations.

To validate the data that AI provides, accountants should start with a general idea of what the outcomes should be based on their assessment of the data before relying on AI. AI should primarily be used to save time and facilitate adapting the information for a presentation.

When generating support materials in a trial, be transparent about using AI. Keep a log of prompts used in generating materials, and be prepared to explain how you validated the output. Because this is a rapidly evolving technology, before using AI for a project, accounting experts must research whether use of AI is permitted in a case and jurisdiction and the extent to which the use needs to be disclosed to other parties.

Here’s a summary of the key points to keep in mind when using AI in accounting expert litigation support:

• Clarify complex concepts on demand. Use AI to translate technical accounting language into clear, accessible terms. Efficiently create summaries that will help attorneys, juries and judges understand complex topics.
• Visualize with precision. Generate charts, timelines and simplified tables to include in briefs and support testimony. AI can create samples you can use or provide a graphic designer tool to improve clarity and comprehension.
• Simulate questions attorneys may ask. Create effective direct-examination questions and anticipate cross-examination by role-playing different legal perspectives. AI can mimic opposing attorneys who might challenge your conclusions, providing helpful expert witness preparation by stress testing explanations.
• Be alert for AI hallucinations and fabrications. AI sometimes “makes things up” with extreme confidence. Always verify technical outputs, references and citations. Avoid relying on AI for authoritative source references without independent confirmation.
• Boost efficiency, not expertise. AI handles basic tasks without replacing expert judgment. Use AI to summarize documents, draft visual aids, and brainstorm explanations and counterarguments. However, your expert judgment should never be derived from the AI itself.
• Remain vigilant to assure confidentiality. Always be vigilant of AI’s legal and ethical implications. Confirm your firm’s AI usage policies before using AI with actual data. Never input sensitive case details, especially evidence under court seal, into any public AI tool. Always use redacted or hypothetical information.

Empowering experts, enhancing justice

AI is more than a tool; it’s a transformative force for accounting experts, especially in litigation support. By simplifying data into effective key point summaries, intuitive charts and clear reports, AI bridges the gap between expert accountants and lawmakers. 

The key to effectively using AI is mastering prompt engineering, balancing AI’s advantages with its drawbacks and remaining focused on the ever-changing ethical and legal parameters associated with it. When adopting AI, it’s critical to have a clear data governance policy, proper training for anyone using AI and an acceptable use policy to make sure AI is employed only when permitted. Accounting experts who embrace AI as a valuable tool will enhance their practice and redefine their role as communicators. 

By streamlining tedious tasks, generative AI tools empower accounting experts to focus more on higher-order functions like persuasive opinion development and courtroom strategy. This evolution in efficiency and capability positions the accounting expert not merely as a technical adviser but as a persuasive force in seeking truth and advancing justice.

Rick Warne, Ph.D., CFE, CPA, is a professor of accounting at the University of San Diego. Contact him at rwarne@sandiego.edu.

Kim Charron, Ph.D., CPA, CMA, is an associate professor of accounting at the University of Nevada, Las Vegas. Contact her at kim.charron@unlv.edu.

Robert Cornell, Ph.D., CMA, is an associate professor of accounting at the University of Nevada, Las Vegas. Contact him at robert.cornell@unlv.edu.