The Changing Nature Of Auditing: Considerations For Incorporating AI

Forbes

21-03-2025

William Tarr is the co-founder and CEO of Tergle, an audit-focused AI software startup in San Francisco.
getty
Seemingly unknown in the world of auditing, the age of AI is dawning. AI-driven financial ecosystems demand new levels of scrutiny and expertise, something that cannot be done with outdated manual methods. Auditing firms must look toward new technologies or risk being dragged into irrelevance.
A century ago, a ledger book, journal, account sheet and slide rule would be sufficient to audit a national public firm. The economy was profoundly more explicable. Steel production saw the processing of a few simple inputs and returned an easily quantifiable output: tonnage of processed metal. Chemicals, similarly, took in industrial scales of unprocessed materials and returned literage of product, while early automobile supply chains could be simply understood and largely remained within national borders.
In the decades after World War II, an era of globalization saw supply chains and monetary transfers become more complex, and new technologies proliferated; the so-called 'Information Age' began. After some adjustment, the tools deployed for the auditing of infant technology companies changed largely in lockstep, with new data processing that arrived in the mid-1980s (Microsoft Excel launched in 1985). The internet and mobile technologies sparked further innovation in the financial services space at the turn of the century. However, as the co-founder of an AI-powered auditing software, it is soberingly clear to me that this is where any meaningful progression stalled.
In my view, failure to adapt has limited the reliability of audits. Even absent the $25 million fine the Public Company Accounting Oversight Board (PCAOB) gave KPMG Netherlands over internal exam cheating, 2024 saw the highest yearly cumulative levels of sanction at $12.4 million in fines to audit firms and individuals. This follows record years in 2023 and 2022 of $11.85 million and $11.02 million respectively, according to the Financial Times (paywall).
Many of these fines have fallen on the 'Big Four' firms, and they highlight broader areas where AI-driven processes can help. For example, in auditing Aegean Marine Petroleum Network, PwC 'ignored red flags' such as a nonexistent address and client businesses reporting addresses in residential blocks, the Financial Times article also said. Such formulaic practices as checking addresses and verifying clients are ripe for automation.
Outside the U.S., the Financial Reporting Council's investigation into PwC and EY in their dealings with London Capital & Finance concluded they had 'failed to understand the business and raised the possibility that there could be 'material misstatements' in the company's accounts,' the Guardian reported. Assurance practices are another area where there's an opportunity for technology innovation.
More egregiously, and something I find perplexing, many audits still rely on data sampling, which can risk sampling errors and incomplete analysis of financial documents. Instead, utilizing LLMs that can collate large sources of information, including learning from peer entities to eliminate any failures to understand business models and conduct whole-scale testing, should be encouraged as a preventative measure.
Those who remain skeptical about the introduction of AI into a field that fundamentally risks liability for material failures abound. It is no surprise, however, given the many years of education required for the field. In response, I want to help allay those fears: In no way do I anticipate the role of auditing to be automated away in some Isaac Asimov-esque future. It would be as if there were 10 very intelligent interns sitting on your shoulder completing repetitive tasks that require little qualitative intervention and flagging discrepancies as and when they arise.
Venture capital-driven bullishness in agentic AI (registration required) has contributed to the proliferation of companies seeking to automate back-office processes. I've noticed more companies (my own included) aiming to bring AI into the auditing profession. AI and data analytics advancements are contributing to the $313 billion market value the auditing services market is expected to achieve by 2031.
Of course, some challenges remain in the implementation of new software, such as the need for high-quality training data. This can be difficult given both the intensity of regulatory control over financial data and the dearth of publicly available audit materials not sourced from large, public corporations. However, I expect these concerns to dissolve as model training becomes more effective and can be bootstrapped through synthetic data generation.
In addition, the past few years have seen AI progress at an unprecedented rate. For instance, optical readers have become up to 99% accurate. There is no reason to say that the current challenges render AI's implementation impossible. The excuses of the past are becoming increasingly hard to defend; the time for AI adoption is now.
Firms seeking to integrate AI should first identify tasks best suited for automation, such as anomaly detection, document processing and risk assessment.
If exploring external AI solutions, it's important to consider whether they can integrate with existing accounting systems, enhance precision and ensure compliance. It's also critical to select AI providers that prioritize transparency, security and explainability—those that enhance, rather than obscure, the auditing process. The firms that move first and modernize workflows while keeping auditors in control will define the profession's future.
There is increasing optimism surrounding the opportunities that exist in agentic AI, and considering the state of auditing, it's easy to understand why. The status quo is untenable. Audit failures, rising fines and outdated methods indicate that traditional auditing can no longer keep pace. Firms that fail to embrace AI could be left behind. Those who cling to the old ways will likely be swept away; in the age of AI, it's adapt or die.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?