How Can Artificial Intelligence Help Tackle Bribery Payments?

In many parts of the world in which commerce is conducted, the difference between a gratuity and a bribe can be difficult to discern. However, the distinction is a critical one. The Foreign Corrupt Practices Act (“FCPA”) – the federal law prohibiting issuers of securities registered on a domestic stock exchange, any company organized under the law of the United States or with its principal place of business in the country and any citizen or resident from bribing foreign government officials to obtain a commercial benefit – makes clear that bribery is a crime with severe repercussions to those that run afoul of its provisions.¹ Nevertheless, it is estimated that the global value of corporate bribery increased to nearly USD1.75 trillion in 2021.²

Furthermore, while corporations are beefing up governance and internal controls to comply with the law and mitigate bribery activity, many of the processes are still manual, onerous and inefficient. When so-called “red flags” do pop up, businesses are often caught flat-footed because they do not have the technology, resources or expertise to address and mitigate potential bribery and corrupt conduct.

Leading companies are adopting the use of artificial intelligence (“AI”) and machine learning (“ML”) as a solution. AI refers to the technology that mimics human intelligence to perform complex, repetitive or time-consuming tasks and iteratively improve as more data is collected. ML is an application of AI. It is the process of applying algorithms (or “models”) to data to help a computer learn without direct instruction. The aim of AI is to replicate and improve the ways that humans analyze datasets and scale up that analysis for efficient and consistent results.

While the use of AI is expanding, the level of adoption for purposes of reducing instances of bribery and corruption still has a long way to go. As with many new technologies, organizations may not know where to start, or fully understand the scope of an implementation or have the right personnel and resources in place. Deploying AI/ML requires organizations to make investments in technology, establish governance processes and facilitate change management, which can be challenging.

Companies that have successfully deployed AI/ML have developed many leading practices. An initial key learning is that AI technology is best implemented when “use cases” or descriptions of how AI can be used to achieve specific goals and tasks is defined, actionable and understood by the organization. For example, a company may wish to use unstructured data (e.g., emails, messages, invoices or other third-party documents) to identify names associated with travel and entertainment (“T&E”) payments and cross-reference them against publicly available data sources to identify potentially improper payments to government officials.

A well-defined use case will facilitate:

Identification of the applicable data (both internal and third party)
Determination of the appropriate model to use (e.g., natural language processing)
Model training and testing approaches to verify that the model operates as defined to a high degree of accuracy
Implementation and monitoring protocols

Case Study

International Investigation of Bribery Activity

Situation

FTI Consulting’s Data & Analytics team built, trained and deployed an AI-driven solution to identify potential improper payments and expenses as part of an internal corporate investigation.

Our Role

To begin, we investigated payment activity using a combination of manual processes and data analysis. We created a set of rules to identify potential bribery activity based on various data attributes and their respective linkages across a procurement system and general ledger. If a linkage between the procurement system and general ledger could be established based upon the rules, the relevant data in each dataset were flagged as linked (i.e., potentially improper).

Our Impact

Once trained, the model we designed and implemented consistently replicated the rules with a high degree of accuracy. It also predicted potentially improper items in new datasets with a high degree of precision. For our client, this exercise served as a successful test case to inform operational decision making around scaling and deployment of AI across the enterprise to augment overall risk management and operations.

Investing in AI Technologies Now Will Reap Long-Term Benefits

As AI becomes more ubiquitous, particularly in its application to combat corporate bribery and corruption, companies and business leaders can learn and apply valuable lessons. To effectively use AI/ML to identify improper payments, it is critical that companies use technology customized for their unique profiles and datasets. An implementation will also benefit if done as part of an overarching change-management plan. When used effectively, the long-term benefits of AI/ML to fight potential bribery can more than offset the initial implementation cost by reducing compliance costs and fostering goodwill, transparency and accountability among all stakeholders.

Footnotes:

1: 15 U.S.C. § 78ff. Penalties

2: “Global Fraud and Risk Report 2021/22 – Research Summary: Bridging the Great Divide,” Kroll (last visited April 19, 2023), https://www.kroll.com/en/insights/publications/global-fraud-and-risk-report-2021/research-summary-bribery-and-corruption