Artificial Intelligence and Anti-Money Laundering: Risks and Rewards

While everyone can agree that artificial intelligence (AI) is the next big thing, the technology is still in its infancy and many industry-specific applications remain somewhat far off on the horizon. One exception is anti-money laundering (AML) compliance, where AI has already shown significant promise and meaningful results.¹

Internationally, regulators are starting to recognize the potential value of AI to drive better AML compliance. On December 3, 2018, major American institutions, including the Board of Governors of the American Federal Reserve System, the Financial Crimes Enforcement Network and the Office of the Comptroller of Currency issued a Joint Statement on Innovative Efforts to Combat Money Laundering and Terrorist Financing (the “Joint Statement”).² The Joint Statement encourages innovation in AI by banks, even going as far as to state that, where banks test or implement AI-based transaction monitoring systems that succeed in identifying suspicious activity that would not otherwise have been identified under existing processes, this will not create an assumption that the bank’s existing processes are deficient or result in additional regulatory expectations.

While there is no equivalent to the Joint Statement in Canada, AI-based AML compliance programs will not be a uniquely American phenomenon. As early as June 2017, HSBC Bank Canada announced it had partnered with a Silicon-Valley-based start-up to automate certain of its compliance processes in an attempt to become more efficient about AML.³ When RBC recently sought to hire a Manager in its AML Risk Analytics group, the bank listed the ability to create effective algorithms and proficiency with natural language processing as “must-have” skills.⁴ The former CEO of TD Bank helped found and is now the chair of Toronto’s Vector Institute, a leading AI research centre.

The urge to apply AI to AML is warranted. Notwithstanding a common recognition of the importance of AML compliance, it has proven difficult for banks and regulators to get ahead of the issue. As reported by Europol, for example, just 10% of suspicious transactions are further investigated after collection; a figure that is unchanged since 2006.⁵ Through machine learning algorithms and other techniques such as frequent pattern mining algorithms, behavioural modelling, risk scoring and anomaly detection, AI provides the opportunity for institutions and regulators to exponentially increase the scale and efficiency of AML detection and compliance programs.⁶ For Instance, IBM Watson Financial Services has suggested that AI-enhanced AML can reduce false positives by 30-50%, and increase decision making time by 30-50%.⁷

At this time, regulators are not promoting any particular AI solution, nor are they penalizing banks for not pursuing more innovative approaches.⁸ While accuracy, consistency and efficiency are paramount, institutions that develop AI solutions at this nascent stage should also seek to reassure regulators that the AI is transparent and accountable. To the extent possible, every variable considered by the AI should be unambiguous and well documented.

While AI provides significant promise in AML in large part because of its potential to scale compliance, increased scale also means increased risk. The characteristics of these risks are familiar to sophisticated counsel; algorithmic bias, for example, is akin to the human biases that have traditionally grounded liability analyses in various legal fora. Goodmans’ Technology Group is naturally attuned to the legal, business and reputational risks that can flow from AI-based AML programs. Institutions and regulators that seek to apply AI to AML should ensure they have carefully considered all potential risks and sought the advice of experienced inside and outside counsel.


¹ See for example, Intel’s “Saffron” AML advisor, which has been deployed by foreign banks: https://www.icc-ccs.org/index.php/1239-ai-money-laundering-and-banks
² https://www.fincen.gov/sites/default/files/2018-12/Joint%20Statement%20on%20Innovation%20Statement%20%28Final%2011-30-18%29_508.pdf
³ https://www.bnnbloomberg.ca/hsbc-partners-with-ai-startup-to-combat-money-laundering-1.767098
⁴ https://ca.linkedin.com/jobs/view/manager-aml-risk-analytics-at-rbc-1176837879
⁵ https://www.europol.europa.eu/sites/default/files/documents/ql-01-17-932-en-c_pf_final.pdf
⁶ https://scinapse.io/papers/2788185337
⁷ https://www.ibm.com/blogs/insights-on-business/banking/category/regtech/
⁸ https://www.ibm.com/blogs/insights-on-business/banking/category/regtech/