Contributed by Shaurya Singh (affiliated with GBLR-SCCLP)
On July 26, 2023, the United States Securities and Exchange Commission (SEC), via a press release proposed a comprehensive set of rules aimed at governing the usage of Predictive Data Analysis (PDA) tools by investment advisers and broker-dealers. This proposal by the SEC underscores a critical concern: that PDA tools, powered by advanced technologies such as Artificial Intelligence (AI) and Machine Learning (ML), can be intentionally designed to prioritize the interests of financial institutions over those of the investors they are entrusted to serve. In response to this pressing issue, the SEC has put forth a set of rules, including record-keeping and disclosure requirements.
This development brings to mind the regulatory landscape in India. The Securities and Exchange Board of India (SEBI) extended disclosure requirements in 2019 to encompass Mutual Funds, Market Intermediaries- which include Stock Brokers and Depository Participants, and Market Infrastructure institutions. However, SEBI is yet to address the issue of potential conflicts of interest arising from the application of advanced AI and ML technologies in investment advisory. This article seeks to provide an analysis of the SEC’s recent proposed rule which directly addresses the issue of conflict of interest exacerbated by AI and ML. It also scrutinizes the effectiveness of the current SEBI regulations governing the use of such technology and presents key takeaways for SEBI’s consideration to protect investors from potential vulnerabilities.
SEC’s proposed rule on usage of AI-ML in Investment Advisory
In its proposed rule, the SEC acknowledged the positive impacts of PDA technologies on capital markets, including enhanced transparency, increased liquidity, and improved overall efficiency. However, the SEC focused on addressing the potential risks associated with the growing use of AI in the realm of investment advisory. The central concern raised by the SEC revolves around the potential for conflicts of interest may emerge as financial firms increasingly incorporate AI into their operations. The SEC’s concern is grounded in the belief that firms may intentionally or unintentionally factor their own interests into the data and software that underlie AI and other PDA-like technologies, ultimately leading to potential harm for investors. This apprehension is further fueled by the observable trend in financial markets, where firms and investors have begun to embrace novel technologies, such as AI, ML, Natural Language Processing (NLP), and chatbot technologies, to drive their investment decisions. For instance, the SEC pointed out that robo-advisors can be intentionally programmed to align with the interests of the institution that deploys them. This alignment might manifest in recommendations that favour the institution’s proprietary products, portfolio rebalancing strategies designed to generate higher fees for the institution, or other recommendations primarily beneficial to the firm itself.
The SEC proposed rules to address conflicts of interest related to AI and ML usage in investment advisory. For investment advisers, these rules necessitate the establishment of written policies and procedures, which include an evaluation process for AI technology used in investor interactions, a method for detecting conflicts of interest, procedures to eliminate or neutralize these conflicts, and an annual review to assess policy adequacy. The SEC also proposes amendments to Rules 17a-3 and 17a-4 under The Securities Exchange Act of 1934 (Exchange Act) and Rule 204-2 under the Investment Advisors Act of 1940 (Advisers Act) mandating documentation and disclosure of AI and ML technologies in investment activities. These amendments encompass documenting evaluations, testing, conflict-of-interest determinations, policies, and disclosures related to covered technologies used in investor interactions. The rules aim to enhance transparency, accountability, and investor protection by requiring comprehensive record-keeping on technologies influencing investment behaviours and outcomes.
AI and ML usage under SEBI’s Oversight
In 2019, SEBI issued three circulars in response to the growing adoption of AI and ML systems in the capital markets. These circulars introduced enhanced reporting requirements for Market Intermediaries, Market Infrastructure Institutions, and Mutual Funds. However, the primary objective behind these circulars was twofold: first, to conduct a comprehensive survey of the AI & ML landscape; and second, to build an inventory of these technologies. The circulars were merely to develop an understanding of AI/ML applications to ensure readiness for potential AI/ML policies in the future and they did not extend their scope to investment advisors.
Nonetheless, SEBI’s Investment Advisor Regulations 2013 (IA Regulations), underscore the core principle of eliminating conflicts of interest in investment advisory services provided to clients. Regulation 15 (1) of the Investment Advisor states that “An investment adviser shall act in a fiduciary capacity towards its clients and shall disclose all conflicts of interests as and when they arise”. The Code of Conduct mentioned in the IA Regulations also conveys that investment advisers should earnestly endeavour to steer clear of conflicts of interest. In cases where avoidance is not entirely feasible, advisers shall make explicit and appropriate disclosures to their clients while ensuring that clients are treated fairly. Additionally, Regulation 19 (f) mandates the disclosure of the rationale behind investment advice to clients. In essence, the IA Regulations mandate that any entity or individual offering investment advisory services must be transparent about both actual and potential conflicts of interest and provide clients with a clear understanding of the reasoning behind their investment advice. Therefore, a prima facie interpretation of the IA Regulations covers the usage of advanced PDA technologies by investment advisors which could lead to a conflict of interest, as the investment advisor is mandated to disclose all the factors that can cause such conflicts.
However, a challenge arises when it comes to the absence of explicit guidelines regarding record-keeping and disclosures, particularly in cases involving opaque technologies such as Black Box Algorithms. These algorithms operate in a manner where their internal mechanisms and decision-making processes remain concealed. As a result, understanding the rationale behind a specific investment decision becomes a complex issue. Hence, Disclosure of such rationale under Regulation 19 of IA Regulations becomes intricate, if not impossible.Surat Stock
In India, Chapter IV along with Schedule II of the SEBI Stock Broker Regulations, 1992, and Chapter III along with the Third Schedule of the SEBI Investment Advisers Regulations, 2013 lay out the record-keeping requirements for broker-dealers and investment advisors which aim to ensure transparent, equitable and fair practices by these intermediaries. However, the existing regulations fall short when it comes to including the inscrutability of black box algorithms under their ambit.
Key Takeaways for SEBI
To address the challenges posed by black box algorithms, the SEC proposes amendments to the Exchange Act and the Advisers Act which lay out a set of disclosure and record-keeping mandates for institutions incorporating AI models, including black box algorithmsAgra Investment. The proposal states that firms should incorporate ‘explainability’ features into their systems and adopt comprehensive back-end controls, including limiting access to the technology and specifying its use cases. This approach ensures that firms can fulfill disclosure requirements even when the inner workings of AI systems remain obscure. The SEC has also maintained that a firm’s lack of visibility into the processing of a form of technology does not exempt it from the obligation to comply with the rules.
Attention may also be given to the approach adopted by the Financial Conduct Authority (FCA) of the United Kingdom. FCA lays out that the firms must define, develop, and monitor trading algorithms comprehensively, aligning with risk appetite and regulatory standards. They need approval processes, pre-post trade controls, real-time monitoring, and compliance checks under the Senior Managers and Certification Regime to ensure algorithmic compliance and oversight. This liability on the senior management is also mandated by the financial regulators of Germany, China, Netherlands, and Luxembourg. This underscores the accountability of firms’ senior management in overseeing AI and ML technologies within trading platforms, emphasizing their responsibility for deployment, testing, and monitoring.
Therefore, to effectively regulate black box algorithms, SEBI should require firms to disclose their use and incorporate explainability features. Additionally, SEBI should emulate the FCA model, holding senior management accountable for algorithmic oversight, pre-post trade controls, real-time monitoring, and approval processes. This approach ensures transparency, accountability, and compliance in the Indian trading landscape.
In an era marked by rapid technological advancements, the potential of PDA technologies to reshape capital market investments and industry norms is undeniable. Yet, the pursuit of this transformation should go hand in hand with vigilant regulatory measures to establish a sustainable and equitable market landscape. The convergence of investor protection, technology-driven innovation, and proactive regulatory foresight becomes imperative to foster a market environment that endures.
While India has taken preliminary regulatory steps to understand these technologies, a comprehensive regulatory framework with explicit guidelines remains a necessity. In this endeavour, SEBI can draw valuable insights from the proactive measures taken by the SEC and the FCA in regulating AI and ML technologies within the realm of investment advisory. By adopting similar initiatives, India can navigate the path to comprehensive regulations and effective oversight, ensuring that PDA technologies not only bring innovation but also uphold fairness and transparency.
Kanpur Wealth Management