Gideon Tsoutsouris

30 December 2024

The emergence of artificial intelligence (AI) has begun to revolutionize every industry on Earth; among the most heavily impacted are banking and capital markets. These technological developments have the potential to create new jobs, revolutionize trading strategy, and alter the stock market. New issues of hyper volatility and trading transparency are cropping up in capital markets, though little has been done thus far to address these concerns. As organizations like the European Union begin to pass legislation to combat these issues, the United States lags behind. In a world where changes are occurring so rapidly, the U.S. government must create regulation that keeps up with artificial intelligence technology in an ever-evolving world for market participants.

A study from the University of Chicago Booth School of Business demonstrates how machine learning technology has already started to prove itself in the trading world. Their research showed that OpenAI’s GPT-4 outperformed human capabilities by about ten percent, with the large language model (LLM) predicting future earnings at a success rate of sixty percent, while humans generally found success in the low fifty percent range [1]. With AI’s results far exceeding human forecasts, there is no doubt that technology will replace human roles and, in combination with AI-driven trading models, will drastically change the stock market of the future.

While artificial intelligence is both more efficient and more lucrative than any human-operated quantitative trading strategies, this radical success brings with it its downsides. AI exchange-traded funds (ETFs) trade at much higher volumes than was previously possible under human management [2], which will likely lead to much more market volatility. This type of volatility could lead to an increase in the frequency of “flash crash” events: instances where the market quickly plummets. If the market is left to experience a new form of volatility like this without the implementation of new financial regulation, investors will most certainly begin to lose confidence in capital markets, which may lead to wider economic impacts.

Given the dilemma that exists between optimizing trading strategies with artificial intelligence and balancing the volatility of the market, regulators must create policy that addresses this issue. The International Monetary Fund suggests a few options for doing so– one being the implementation of new “volatility response mechanisms” [2]. Because of the risk of “flash crash” events associated with AI-driven trading, implementing mechanisms such as AI-related circuit breakers can stop these from occurring. Circuit breakers stop trading to prevent panic selling in the event that prices drastically drop in a short period of time.

Another major concern regarding AI in capital markets is the degree of transparency that corporations are required to maintain. The IMF suggests that banks and other financial institutions should be required to disclose whether they are using artificial intelligence in their trading, so as to keep others engaging with the market informed on which firms are using this new technology. This could help anticipate sudden drops in the market  and provide an understanding of why the market may be moving the way it is in any given situation. A bill requiring this type of transparency with regard to artificial intelligence in America’s financial institutions has the potential to ensure that the stock market is a fair environment for all traders.

In an effort to promote legislation on artificial intelligence in the United States, a bipartisan group of senators led by Chuck Schumer released a report this year that includes ideas on how to increase the transparency and explainability of AI in the financial sector in future legislation. It recommends that legislation be passed requiring transparency of the data used in AI models [3]. However, the United States Senate still has yet to officially propose a bill addressing some of the concerns brought up in this report. While this “roadmap” is not actual legislation, it is a step in the right direction for regulation in the U.S., and a sign that the government is aware these issues must be addressed in the near future.

Across the Atlantic, the European Union took care to address this issue in a concrete piece of legislation passed in March of this year. The EU’s AI Act includes regulation for financial institutions, in part differentiating between types of AI, classifying some as “high-risk.” Such AI systems have increased transparency requirements under this legislation. Similarly to what was recommended by U.S. Senators, EU regulators and government organizations will have the power to access a firm’s AI system’s training, datasets, and source-code. This will allow governments to investigate the compliance of high risk artificial intelligence with the EU’s act and make others outside of firms aware of where AI systems are deriving their outcomes. Failing to comply with these regulations comes with tens of millions of euros in fines [4].

Another major organization that took steps in order to regulate artificial intelligence was the International Organization of Securities Commissions. IOSCO includes “ninety-five percent of the world’s securities markets,” [5] making it one of the most influential regulatory associations of capital markets in the world. They hold similar reasoning to the EU as to why they recommend regulation that promotes transparency surrounding AI-driven strategies. IOSCO wants to make sure that artificial intelligence algorithms are explainable and transparent in order to maintain an understanding of the workings of capital markets in the event that corrections need to be made. They do not want firms or governments to be operating blindly when utilizing AI in capital markets.

Since the United States’ Securities and Exchange Commission does belong to IOSCO, the SEC is not directly subject to their regulations, but is simply a participating member in creating policy recommendations. The SEC, along with so many others, have proposed rules for AI usage in capital markets, with emphases on disclosures, compliance programs, and data protection [6]. A specific area of concern, which the SEC addresses in their recommendations, is the risk of what they call AI “monoculture.” If too many firms use the same AI-driven datasets or models for their trades, it is more likely that the trades made by this software will be increasingly similar with competing firms, concentrating risk in certain areas of the market. In order to prevent this high concentration of risk, which could put the market in danger of becoming even more volatile, the SEC has proposed rules which would ban firms from using “predictive data analytics” in ways that take the firm’s interests into account over those of their clients [6].

While organizations like the Financial Industry Regulatory Authority (FINRA), the SEC, and the United States Senate have suggested measures for the regulation of AI, the U.S. does not currently have any federal regulation in place for artificial intelligence’s use in the purchases and sales of securities [7]. This must change soon, as the European Union and international regulatory organizations have already adopted such policy. Artificial intelligence is not going anywhere, so the government in the United States needs to be prepared for changes in capital markets and be ready to address the risks associated with AI in such a consequential area of financial regulation. Whether this policy comes from regulations put into place by the SEC, or Congress is able to pass a bill mandating similar regulations, it is necessary for the federal government to step in and protect capital markets and the financial system.

---

Image via Pexels Free Photos

Works Cited

[1] OECD (2024), “Regulatory approaches to Artificial Intelligence in finance”, OECD Artificial Intelligence Papers, No. 24, OECD Publishing, Paris, https://doi.org/10.1787/f1498c02-en.

[2] Abbas, Nassira, Charles Cohen, Dirk Jan Grolleman, and Benjamin Mosk. “Artificial Intelligence Can Make Markets More Efficient-and More Volatile.” IMF, October 15, 2024. https://www.imf.org/en/Blogs/Articles/2024/10/15/artificial-intelligence-can-make-markets-more-efficient-and-more-volatile\.

[3] Michael, David A. Simon, Merin P. Cherian, Daniel,. “Understanding SEC’s Focus amid Lack of Final AI Rules: Insights: Skadden, Arps, Slate, Meagher & Flom LLP.” Insights | Skadden, Arps, Slate, Meagher & Flom LLP, February 23, 2024. https://www.skadden.com/insights/publications/2024/02/understanding-sec-focus-amid-lack-of-final-ai-rules. 

[4] “Fact Sheet.” IOSCO, November 2022. https://www.iosco.org/about/pdf/iosco-fact-sheet.pdf.

[5] Pierides, Mike, James Mulligan, and Christopher Archer. “Henrystewartpublications.” Journal of Securities Operations & Custody, May 9, 2022. https://www.henrystewartpublications.com/sites/default/files/JSO14.4AI%20in%20capital%20markets%20Navigating%20legal%20and%20regulatory%20considerations.pdf.

[6] “REGULATORY APPROACHES TO ARTIFICIAL INTELLIGENCE IN FINANCE.” OECD Publishing, September 2024.

[7] Kelly, Jack. “Ai Could Displace More than 50% of Banking Jobs, According to New Citigroup Report.” Forbes, August 27, 2024. https://www.forbes.com/sites/jackkelly/2024/06/20/ai-could-displace-more-than-50-of-banking-jobs-according-to-new-citigroup-report/.

[8] Clarke, Paul. “‘Revenge of the Liberal Arts Graduates’: How AI Will Upend Dealmaker Jobs.” Financial News, March 15, 2024. https://www.fnlondon.com/articles/ai-banking-jobs-citi-goldman-chatgpt-20240415.

[9]“The Impact of Artificial Intelligence on the Banking And …” Deloitte, 2024. https://www2.deloitte.com/content/dam/Deloitte/cn/Documents/financial-services/deloitte-cn-fsi-ai-in-banking-en-240805.pdf. 

[10] Lopate, Lily. “Nvidia GTC 2024: How AI and HPC Are Creating a ‘New Frontier of Capital Markets.’” Technology Solutions That Drive Business, May 5, 2023. https://biztechmagazine.com/article/2024/03/nvidia-gtc-2024-how-ai-and-hpc-are-creating-new-frontier-capital-markets.

[11] “IBM Study: Banking and Financial Markets CEOS Are Betting on Generative AI to Stay Competitive, yet Workforce and Culture Challenges Persist.” IBM Newsroom, June 5, 2024. https://newsroom.ibm.com/2024-06-05-IBM-Study-Banking-and-Financial-Markets-CEOs-are-betting-on-generative-AI-to-stay-competitive,-yet-workforce-and-culture-challenges-persist.

[12] Bi, Shuochen, Wenqing Bao, Jiangshan Wang, and Tingting Deng. Application and practice of AI technology in quantitative investment, n.d. Accessed November 19, 2024. 

[13] Morales, James. “Can Numerai Succeed Where Others Failed? A Look at Crowdsourced AI in Quant Trading.” Edited by Samantha Dunn. CCN.com, September 26, 2024. https://www.ccn.com/news/technology/numerai-succeed-where-quantopian-failed/. 

[14] Patel, Vinay. “Are Financial Analysts at Risk? AI Outperforms in Financial Forecasts and Delivers Superior Trading Profits, Study.” International Business Times UK, May 29, 2024. https://www.ibtimes.co.uk/are-financial-analysts-risk-ai-outperforms-financial-forecasts-deli vers-superior-trading-1724812. 

[15] Director, Shaun HurstTechnical. “Regulatory Alert: Navigating the EU AI Act for Financial Services Firms.” Smarsh, August 1, 2024. https://www.smarsh.com/blog/thought-leadership/regulatory-update-navigating-eu-ai-act-f or-financial-services-firms

[16] Impact-of-the-new-EU-ai-regulation-on-financial-sector- …, September 2021. https://www.cliffordchance.com/content/dam/cliffordchance/briefings/2021/09/impact-of-th e-new-eu-ai-regulation-on-financial-sector-firms.pdf.

[17] “Responsible Artificial Intelligence in Financial Markets: Opportunities, Risks, and Recommendations.” Artificial Intelligence in Financial Markets, May 2, 2024. file:///Users/gideon/Downloads/TAC_AIReport050224.pdf

[18] Donegan, Thomas, David Wakeling, and Hillary Sunghee Seo. “Zooming in on Ai – #5: AI under Financial Regulations in the U.S., EU and U.K. – A Comparative Assessment of the Current State of Play: Part 1.” A&O Shearman, September 23, 2024. https://www.aoshearman.com/en/insights/ao-shearman-on-tech/zooming-in-5-ai-under-financial-regulations-in-the-us-eu-and-uk-a-comparative-assessment-part-1.