The scale and depth of organised financial crime is vast. Figures from the National Crime Agency (FCA) in the UK claim up to £90 billion is laundered in the UK each year, but that number could be significantly underestimated. Drawn to the relative simplicity in registering businesses, the thriving property market, the cachet and credibility of doing business in the UK and its position as a financial hub, the nation has become an attractive target. Globally, the amount of money laundered annually has been estimated at up to 5% of global GDP, or up to $2 trillion.
Organisations including banks, insurance companies and online retailers are constantly on the hunt for the tools, technology and talent to help them fight these crimes. Meanwhile, fraudsters are becoming more sophisticated, their crimes more complex and hidden, and the values ever-increasing as they find ways of circumventing systems.
The most critical element in an organisation’s anti-money laundering (AML) armoury is knowing who they are doing business with – in fact, 84% of businesses say that if they were certain about a customer’s identity, the need for fraud risk mitigation would be reduced.
Now, financial organisations and enterprises are looking to software which incorporates techniques such as natural language processing (NLP), to give them transparency and help detect, locate and prevent the intricate web of financial crimes. NLP is helping enterprises and financial services firms stay a step ahead in the fight against organised financial crime.
Organisations are using to answer questions, to match customers against ‘caution lists’, and to gain deeper levels of understanding. This gives them better visibility into who they’re doing business with, to verify and validate customer details and to confirm identities.
The surging NLP market
NLP isn’t new – it’s been around for about 50 years –but the explosion in data generation has brought it to the fore, leading to investment and improvement in the technology. The global NLP market is forecast to rise to a value of $16bn by 2021.
Facebook recently bought NLP start up Bloomsbury AI (in a bid, it was reported, to combat fake news) and in 2016, Google bought API. AI, a company which built tools for natural language understanding across web applications and mobile devices. Now, enterprises are taking advantage of improvements in NLP to help them shed light on their clients.
98% of red flags are false positives
With financial institutions facing unprecedented fines for non-compliance, it’s no surprise that they are erring on the side of caution when it comes to financial crime investigations – the one instance they miss could be enough to bring down the entire organisation, or worse. However, exercising this air of caution has its downside.
When a transaction monitoring system identifies a ‘red flag’, or unusual transaction which doesn’t fit a standard pattern, employees must painstakingly trawl through these transactions, behaviours and entities manually. One bank cited in an Aite Group report employs 5,000 people for the specific purpose of investigating these suspicious activity reports also referred to as suspicious transaction reports. In reality, 98% of these flags turn out to be ‘false positives’. Organisations know that most businesses and transactions are perfectly legitimate, but they cannot take any risks.
“Regulators expect financial institutions to find every needle in the haystack — false-negatives are not acceptable. This expectation leads to an abundance of false-positives in many current solutions,” said Julie Conroy, research director of Aite Group.
As well as consuming valuable resource and potentially resulting in multi-million pound fines, false positives can have a devastating impact at a personal level, on trust, on the customer experience and on company reputation – a business may find its business account frozen or closed down.
Intelligent systems understanding natural language
Financial institutions have been looking at different ways of accelerating the investigations into these false positives, shifting from manual, rules-based processes to smarter systems incorporating machine learning, and this is where NLP comes into its own.
NLP can read, process and derive meaning from huge volumes of unstructured data in sentence format which may be relevant to a fraud investigation, but previously hidden – email, news articles and social media posts, for example. With NLP, it isn’t just about content, but context. Certain content might, taken alone, ring alarm bells.
When viewed in context, it means something entirely different. NLP applies this logic to its processing, taking context into account. This helps whittle down incidences which may previously have been identified as fraudulent. In financial crimes compliance and anti-money laundering, NLP reads new sources to find mentions of suspects or ‘bad actors’ and understands what those sources are saying about the people mentioned.
Financial services companies are using screening solutions integrating NLP to help them uncover relevant information that would otherwise remain unseen, then match it to the flagged records and against federally-monitored caution lists from sanctioning bodies, law enforcement agencies and regulators, the politically exposed persons (PEPs) list and internal lists.
Screening solutions link, detect and explore relationships, and find transactional patterns that reveal behaviours. With this data, they can perform data quality checks and entity resolution, then match it, cutting investigation time significantly, reducing false positive detection and generating millions of pounds’ worth of savings.
Using screening solutions which integrate NLP, financial institutions divert their investigations to the most significant cases, and gain transparency across their organisation – and those of their clients. By reducing false positives, they are improving the customer experience, building customer trust and playing a crucial role in the global fight against organised financial crime.