Reducing Audit Risk Through Artificial Intelligence and Automation
Publicly listed companies must comply with complex accounting and investor-protection regulations, and that landscape is not static. Governing authorities are continually adding regulations while also updating existing ones. That creates a huge task for organizations managing compliance in a single country. For multinational companies, the work becomes exponentially harder. The challenges of verifying and complying with numerous, complex regulations are ultimately passed on to audit and assurance firms.
Noncompliance can have serious consequences — way beyond the company — as processes become increasingly connected and functions both inside and outside the enterprise become interdependent. This means the responsibility for ensuring compliance is that much greater for auditors. Recently, several audit firms appeared in court to justify nonreporting of client irregularities and failure to detect fraud. To manage these types of liabilities, auditing firms are now deploying software throughout the audit life cycle. This gives auditors additional tools to achieve the following:
- Expand their scope
- Identify high-risk transactions.
- Detect anomalies and fraud in financial reporting.
- Optimize resources.
- Increase margins.
- Protect their brand image and that of their clients.
Optimizing audit and assurance
Audit firms are working with technology companies to explore the use of robotic process automation, machine learning, data analytics and artificial intelligence. These technologies automate repetitive manual tasks, identify hidden patterns of fraud and locate scenarios where mandatory processes have been circumvented. Auditors identify risks and controls in their clients’ daily operations and then identify control objectives, strategic objectives, and frequency and type of risk. From the foundational stage of data integration and analytics through the advanced stage of cognitive intelligence, technology is helping audit firms in all areas.
Technology is optimizing the audit cycle, increasing speed and accuracy, reducing costs, and ensuring efficient deployment of auditors. Some audit processes that can benefit from the new technologies include:
- Assessing engagement risks and negotiating deals — Data analytics can not only help identify client engagement risks but also narrow the scope of work and properly estimate the effort required. This reduces the risk of cost overruns.
- Planning audits — Large audits require detailed plans for timelines, resources, scope coverage, location coverage, entity coverage, consolidation work, communication strategy, and risk and mitigation. Data analytics and RPA help finalize the plan and provide a clear vision of the entire audit process and required milestones. Any flaw in the plan can result in the loss of an important material coverage area. Data analytics help identify what is to be covered, while RPA can direct the planning.
- Performing audit fieldwork — Analysis and forensics often take a back seat in the current audit process, which is largely manual and focused on extensive data collection. With automation, forensics becomes more important and reduces the risk of noncompliance. AI-powered semantic intelligence compiles large quantities of data and automatically recommends enhancements and remediation of important controls. Control failures are detected in near real time, which allows auditors to flag concerns and enables enterprises to take corrective action before an issue reaches regulators.
- Identifying exceptional behavior — With machine learning in place, auditors can identify exceptional behavior in contracts and measure the impacts. They can also plug loopholes that are prone to fraud.
- Generating reports and analysis — RPA, combined with analytics and natural language processing, can help generate reports across verticals, locations and entities. It can also provide detailed reports with a user-friendly presentation and highlight areas that need attention.
While AI offers great opportunities, organizations must consider the associated risks. One of these dangers is universal for AI: embedding human biases in the algorithms.
Amazon dropped experimental recruiting technology, powered by machine learning, last year when the online giant discovered the tool was biased against women. AI has also led to online image searches that ranged from racially insensitive to outright racist.
These problems aren’t new. As noted by Harvard Business Review, the UK Commission for Racial Equality concluded that a British medical school’s automated system discriminated against applicants who were women or had non-European names. That incident was in 1988.
In a report on AI, the Institute of Internal Auditors listed bias as one of the important risks that must be considered. The association also warned about the risks of embedding human logic errors, inadequate testing and oversight, and possible financial or reputational damage.
Organizations must carefully consider the use of AI in auditing, with an appreciation of the technology’s limitations.
Even with such great advances, human intervention is required to complete an audit. At CFO.com, Brian Peccarelli, co-chief operating officer of Thomson Reuters, pointed to the collectability of accounts receivable and the valuation of goodwill and other intangibles as areas where humans will continue to take center stage.
However, digital technology has evolved in a way that significantly reduces reporting risks. Auditing firms can offer better-quality service, optimize resources, expand coverage, detect flaws early in the process and easily track the complete audit process.