How Entity Extraction Protects against Corrupt Actors

A Tale of Financial Corruption

Back in 2018 the then prime minister of Malaysia, Najib Razak, lost a general election. Aside from not being good for his career, the loss led to potentially catastrophic legal consequences for him: shortly thereafter, deprived of his de facto immunity due to holding national office, Malaysian authorities raided his properties and both he and his wife were charged with corruption. The charges mainly focused on alleged embezzlement from the multibillion dollar state investment fund called 1MDB.

Just coincidentally, at the same time when large sums of money had gone missing from 1MDB, his personal bank accounts magically received hundreds of millions of dollars in deposits. They came mostly from mysterious front companies.

Financial Institutions Need to Be Careful in Dealing with Politically Exposed Persons

Mr. Najib is the perfect poster boy for a Politically Exposed Person (PEP). A PEP is a high-level (or in many cases, medium-level) public person in a foreign country that financial institutions must watch very closely. They are obliged to distinguish the honest ones from those who may be corrupt. If they don’t, and if they end up doing business with one of the latter, a financial institution risks being the target of liability suits as well as suffering reputational loss and being exposed to criminal liability.

As a consequence, financial institutions constantly monitor transactions of PEPs to see if there are signs of someone laundering embezzled wealth. (This occurs even if authorities have not uncovered any evidence of criminal activity yet.) PEPs who are living beyond their documented means may well raise red flags as well as eyebrows. To counter this kind of corruption, financial institutions can determine if there are legal limits on items like salary in the foreign country and include them in the transaction monitoring they are obliged to do.

How Entity Extraction Supports Keeping Tabs on Bad Actors

An extremely valuable source of adverse data about PEPs is contained in news media of all kinds. They report on many aspects of a PEP’s personal and professional life. However, it’s obviously difficult for a financial institution to monitor such a massive amount of material.

Here’s where an AI-based technology called Entity Extraction comes in. It can find data about PEPs in unstructured media sources in a scalable, real-time manner. The basic level of Entity Extraction, also known as Named Entity Extraction, automatically identifies semantic concepts in text such as named entities (“Najib Razaq”). Advanced Entity Extraction goes beyond just named entities and identifies more complex semantic concepts such as relationships and events that those individuals participate in.

Entity Extraction Identifies Not Just Names but also the Activities and Associations a PEP Participates in

Event Extraction, which we have discussed in another blog, is particularly valuable in collecting adverse information about PEPs because it identifies business, legal, criminal, or other kind of events in which the PEP is involved. It transforms the unstructured data containing this information into a predictable format that allows easy aggregation.

Relationship Extraction, likewise discussed elsewhere, is also critical, as it identifies relatives and business associates of a PEP. In many cases, the individuals who are directly involved in corrupt activities are relatives acting on behalf of a PEP who remains behind the scenes.

In sum, Entity Extraction provides sophisticated Adverse Media Monitoring to catch the PEPs who are engaged in corrupt activities.