Artificial intelligence (AI) is transforming the global financial services industry, including by helping financial institutions offer innovative new products, increase revenue through efficiencies, and improve customer service. Within the securities and commodities industry, AI-based applications are advancing the sector in customer interactions, investment and trading processes, market surveillance, and operational functions. In order to operate, AI technology needs three components: data, algorithms, and human interaction.
AI APPLICATIONS IN THE SECURITIES AND COMMODITIES INDUSTRY
Customer Interactions: Virtual assistants can be programmed to perform simple digital customer service tasks, including keeping track of account balances and portfolio holdings, market data, address changes, and password resets. Other functions include screening and clarifying income, sending client emails, and performing targeted outreach to customers based on their investing behaviors.
With any widespread use of technology, there are a number of issues to keep in mind, including how to maintain customer privacy, eliminate bias in programming, and avoid instances where the technology is used by actors to commit fraud. Other issues to keep in mind are the customer authentication process, cybersecurity needs, and fair and accurate recordkeeping.
Account Management: Customer profiles can be created and analyzed based on their assets, held both at the investment firm and externally, as well as their spending patterns, debt balances obtained through data aggregation tools, updates on social media and other public websites, browsing history on the firm’s website and mobile apps, and past communications. AI-based tools also may provide customers’ social media data and related sentiment analysis on investment products and asset classes.
Portfolio Management: New patterns can be identified, potential price movements of specific products or asset classes can be predicted, and satellite activity can be interpreted to improve portfolio management. A significant trend over the last decade has been the introduction of automated advisers—roboadvisers—which use algorithms to provide advisory services over the internet.
Trading: AI can assist with smart order routing, price optimization, best execution, and optimal allocations of block trades, in addition to automated algorithmic trading. With a number of proposed rules and open comment periods from the US Securities and Exchange Commission (SEC), any rulemaking will have the potential to generate significant data sets for the SEC to use in monitoring the markets and market participants.
Surveillance and Monitoring: AI can capture and surveil large amounts of structured and unstructured data in various forms, such as text, speech, voice, image, and video data, from both internal and external sources, in order to identify patterns and anomalies. AI has the ability to decipher tone, slang, and code words. To improve market surveillance, AI could be used for predictive, risk-based surveillance.
Know-Your-Customer and Customer Monitoring: Machine learning, natural language processing, and biometric technologies could be implemented to detect potential money laundering, terrorist financing, bribery, tax evasion, insider trading, market manipulation, and other fraudulent or illegal activities that continue to be threats to the industry.
Regulatory Intelligence: New and existing regulatory intelligence can be digitized, reviewed, and interpreted, including rules, regulations, enforcement actions, and no-action letters, and appropriate changes can be incorporated into compliance programs. Regtech is on the rise, as demonstrated by the Financial Industry Regulatory Authority (FINRA) initiative to provide a machine-readable rulebook.
Liquidity and Cash Management: AI systems can be used to identify trends, note anomalies, and make predictions; for example, related to intra-day liquidity needs, peak liquidity demands, and working capital requirements.
Credit Risk Management: AI systems can be used to provide more accurate and fair credit risk assessments by retrieving troves of data not used in traditional credit reports, including personal cash flow, payment applications usage, on-time utility payments, and other data buried within large datasets.
REGULATORY USES, CONSIDERATIONS, AND PERSPECTIVES
Financial regulators are increasingly turning to AI to enhance and streamline their processes and systems. Through technological advancements, regulators have more efficient monitoring methods and the ability to collect wider ranges of data sets, perform more extensive analysis, and make compliance more cost-effective for financial institutions.
The use of AI is changing the regulatory landscape from that of a static, rule-based one into a dynamic, risk-based paradigm.
Launched in October 2022, the FINRA Rulebook Search Tool (FIRST) is a machine-readable rulebook through the creation of an embedded taxonomy—a method of classifying and categorizing a hierarchy of key terms and concepts—that was applied or “tagged” to the 40 most frequently viewed FINRA rules, allowing users to narrow down the universe of potentially applicable rules through sophisticated search filters. The comment period runs through February 21, 2023.
When it comes to FINRA’s exam priorities, AI can review disclosures, complaints, or employment history data to help staff determine which registered representatives to examine.
FINRA has begun using deep learning for market manipulation surveillance to address changing market conditions, increased volatility, increased volumes, and change in conduct in order to protect investors and ensure market integrity. Working closely with the SEC and the securities exchanges, FINRA plays “a central role in conducting ongoing oversight within and across markets, monitoring for misconduct and intervening promptly” once discovered. By being able to react faster, FINRA believes it is using deep learning to make the market safer.
In July 2022, the Commodity Futures Trading Commission (CFTC) announced that LabCFTC, a unit focused on “efforts to promote responsible fintech innovation and fair competition,” will be restructured to “take on a new identity” as the Office of Technology Innovation (OTI) and serve as the CFTC’s “financial technology innovation hub, driving change and enhancing knowledge through innovation, consulting/collaboration, and education.”
As a market regulator, the CFTC could leverage AI to distinguish salient activity, use data to develop market models, and identify risk factors.
In December 2020, the CFTC adopted a final rule addressing electronic trading risk principles, marking a shift toward a principles-based approach to regulating automated traded compared to the CFTC’s previous regulatory efforts.
Without any official guidance, financial agencies likely will regulate AI by enforcement. The CFTC has brought several cases involving spoofing, and the SEC has brought enforcement actions involving governance over an investment model’s algorithm and against digital advisers for misleading disclosures in marketing materials.
CONSIDERATIONS WHEN BUILDING AN AI COMPLIANCE PROGRAM
While various organizations have proposed frameworks for AI, an investment firm has some flexibility in creating an AI compliance framework. Some frameworks use guiding principles that include governance data, performance, and monitoring.
When addressing how to build an AI compliance program, a firm should conduct an inventory of existing AI systems; assess where future AI systems will be used; evaluate existing or establish new AI-specific policies; assign responsibility for or designate a role to handle AI initiatives and ongoing monitoring; keep records, including of third-party systems; and be prepared to respond to regulatory inquiries or otherwise discuss AI systems with regulators.
Watch our on-demand Artificial Intelligence Boot Camp session for more information on AI’s role in the securities and commodities industry.