TL;DR:
- Acting Comptroller Hsu discusses the benefits and risks of AI and tokenization in banking, emphasizing the need for collaboration between banks and regulators.
- Hsu highlights challenges such as alignment issues, discrimination and bias, and the potential for AI-enabled fraud and misinformation.
- Google Cloud launches an AI-powered anti-money laundering program, showing promising results in detecting suspicious activities.
- Hsu expresses skepticism about the crypto industry’s maturity and emphasizes the potential of blockchain tokenization to improve settlement efficiency.
- Hsu proposes a prudent approach to innovation, including incremental stages, involving risk professionals from the start and engaging regulators throughout the process.
Main AI News:
The growing landscape of artificial intelligence (AI) and tokenization in the banking industry has prompted Acting Comptroller Michael J. Hsu to address the risks and benefits associated with these innovative technologies. In his recent remarks at the American Bankers Association (ABA) Risk and Compliance Conference in San Antonio, Texas, Hsu emphasized the need for banks to collaborate closely with regulators to effectively manage the inherent technological risks.
Hsu’s timely remarks coincided with an announcement from Google Cloud, which unveiled an AI-powered anti-money laundering program just five days later. While the initial results of Google’s program seem promising, only time will reveal whether Hsu’s concerns regarding the risks of AI will prove prophetic.
AI Adoption and Challenges Hsu acknowledged that banks had approached the adoption of machine learning and AI with caution. Despite the market’s enthusiasm for AI’s rapid growth, Hsu noted that media reports often exaggerate the extent of AI adoption in banks, leading to speculative interpretations rather than an accurate reflection of reality.
For banking institutions, Hsu stressed that both the risks and benefits of widespread AI adoption are substantial. Among the risks, Hsu identified the challenge of aligning AI systems with banks’ objectives and values. Unlike traditional software, AI requires training and may not always behave as intended. This misalignment can give rise to governance and accountability challenges, creating opportunities for plausible deniability when things go wrong. Hsu also raised the question of accountability, specifically whether the bank or its third-party AI vendor should be held responsible for AI-related failures.
Furthermore, Hsu highlighted the unique discrimination and bias challenges posed by AI. Even if an AI system achieves color-blindness at the individual level, disparities can still emerge at the group level if underlying baselines differ. Hsu also cautioned against the potential for AI to enable fraud and the spread of misinformation. The speed and sophistication of AI’s development, combined with its ability to mimic human communication, could significantly increase opportunities for fraudulent activities, requiring close monitoring and coordination from banks and regulators.
Hsu underscored the importance of addressing these challenges, particularly regarding the role of social media in facilitating the spread of miscommunication. Citing an example of a fake Bloomberg Twitter account posting misleading information, which triggered market turbulence, Hsu emphasized the need for banks and regulators to update their playbooks and strengthen defenses against such actions.
Despite recognizing the potential advantages of AI, Hsu’s remarks conveyed an overall tone of caution and skepticism, urging a measured and careful approach. Throughout his speech, he emphasized the importance of implementing adequate “brakes” in the AI adoption process.
AML AI in Action In line with Hsu’s remarks, Google Cloud recently launched Anti Money Laundering AI (AML AI) on June 21, 2023. AML AI is an AI-powered product designed to enhance financial institutions’ ability to detect money laundering and terrorist financing efficiently. It leverages machine learning-generated customer risk scores, drawing on transactional patterns, network behavior, and Know Your Customer (KYC) data from the bank. This approach offers a consolidated alternative to rules-based transaction monitoring, outperforming existing systems in identifying suspicious activities.
HSBC Bank, a Google Cloud customer, reported significant improvements in their AML detection capabilities through AML AI. The bank experienced a notable increase in “true positives” while reducing transaction monitoring alert volumes by over sixty percent. HSBC’s Group Head of Financial Crime Risk and Compliance credited Google Cloud’s sophisticated AI-based product for enhancing their financial crime detection precision, reducing investigation time spent on false leads, and shortening the analysis period for billions of transactions.
Tokenization and Settlement Efficiency Hsu expressed skepticism about the crypto industry, highlighting its immaturity and associated risks. Fraud losses, scams, and hacks in the sector have reached alarming figures, exceeding billions of dollars. Recognizing these risks, regulatory bodies, including the OCC, Federal Reserve, and FDIC, have issued statements emphasizing the importance of risk management for banks involved in crypto activities.
While Hsu remains cautious about blockchain technology, he believes it holds the potential to significantly improve settlement efficiency through the tokenization of real-world assets and liabilities on trusted blockchains. Hsu explained that current settlement processes often involve multiple steps and entities, leading to delays, increased risks, and higher costs. Tokenization has the potential to streamline these processes by collapsing them into a single step, thereby minimizing risks associated with reconciliation and verification. For tokenization to succeed, however, Hsu emphasized the need for legal foundations and regulatory frameworks that support innovation while addressing controls and risk management.
A Prudent Approach to Innovation Against this backdrop, Hsu proposed a prudent approach to innovation in the banking industry, particularly regarding AI and tokenization. He outlined three key principles: innovating in stages, building the brakes alongside the engine, and engaging regulators early and frequently.
Innovation should be pursued incrementally, with disciplined expansion and careful monitoring. Banks should start with what they can control, gradually expand their initiatives, and iterate the process based on monitoring and adjustments. Hsu emphasized the importance of robust new product approval processes and due diligence before embarking on new activities.
To build the brakes while building the engine, Hsu stressed the need to involve compliance and risk professionals from the inception of new product development. By incorporating risk input from the outset, banks can mitigate the financial, legal, and reputational costs that often arise after a product’s launch. Hsu emphasized the value of giving risk and compliance professionals a seat at the innovation table to ensure a well-balanced approach.
Lastly, Hsu underlined the significance of engaging regulators throughout the innovation journey. However, he acknowledged that regulators must possess the necessary agility, responsiveness, and knowledge to effectively support and oversee technological advancements. To that end, Hsu’s office expanded and upgraded its Office of Innovation to the Office of Financial Technology, appointing a Chief Financial Technology Officer to enhance regulatory capabilities in the rapidly evolving landscape.
Conclusion:
The insights shared by Acting Comptroller Hsu shed light on the evolving landscape of AI and tokenization in the banking industry. While acknowledging the benefits, Hsu’s cautious tone reflects the importance of addressing challenges such as alignment, discrimination, and fraud risks. The launch of Google Cloud’s anti-money laundering program highlights the potential of AI to enhance detection capabilities. Hsu’s skepticism towards the crypto industry underscores the need for risk management. Overall, banks and regulators must navigate this dynamic landscape prudently, embracing innovation while mitigating associated risks, to ensure sustainable growth and resilience in the market.
