The Role of Artificial Intelligence and Machine Learning in Sanctions Compliance



Global sanctions regimes are getting more and more challenging for compliance officers to successfully navigate. Is leveraging AI and ML technologies the answer?

Over the past 10 years, trade and economic sanctions have seemingly become the most frequently employed tool of foreign policy. Aside from country-specific sanctions, such as those against Russia and North Korea, there are now more targeted regulations, focusing upon particular businesses or individuals. As a result, anti-money laundering (AML) screening obligations at both national and global levels have increased in both scope and complexity.

The pressure to find a reliable solution to minimize or even eliminate those risks is more intense now than ever.

The limitations of traditional technologies

One of the consequences of the growing complexity of compliance regimes is the inability of traditional technological approaches to combat the evolving threats. Financial institutions are facing a substantial rise in false positive hits.

According to a 2018 Reuters investigative report , for some organizations it’s as many as 95% of their hits prove to be false. The need to manually go over and review this mass places considerable operational and cost burdens on those businesses and institutions.

Many of the systems in the market today use the traditional rule-based approach that can’t assist compliance professionals in weeding out the rising proportion of false positives in our constantly changing world.

How can AI and ML aid in compliance

More recently, though, organisations have started to look to artificial intelligence (AI), natural language processing (NLP) and machine learning (ML) to bolster the accuracy of financial crime detection.

Let’s take a quick look at just a few of the applications of AI/ML that AML compliance could stand to benefit from:

  1. Self-enriching databases
    ML-powered screening engines like DataSpike learn from the searches and enrich themselves with new data continuously. It means they have the ability to proactively adapt to new regulatory expectations.

  2. Spotting data anomalies
    AI models can be trained to identify data anomalies and relationships amongst suspicious individuals and entities.

  3. Advanced name screening
    Rule-based text matching is not an effective tool to make sense of nuances such as the order of names, titles, salutations, abbreviations, name variants, common misspellings, etc. ML algorithms do this with ease, however, striking the right balance between accuracy and the fuzziness of presented data.

As you can tell, leveraging the power of AI/ML for sanctions screening produces a sharp reduction in false positives, which in turn reduces the cost of extra manual labour required for verification. Not to mention the vast improvement to the level of risk of being fined that comes with letting a true positive slip through the cracks of a legacy-based system.

By using an AI-powered approach, businesses, banks and other financial institutions can overcome today’s increasingly complex sanctions challenges. There is no better time than now to review your existing sanctions screening processes and demand quick and proactive adaptation of new technologies to stay ahead of the curve.