The stability of the U.S. financial system is a key pillar of global economic stability and international markets. However, ongoing technical and technological advancements and current geopolitical transformations have introduced new threats to this financial system, including: attacks on financial trading models, a flooding of U.S. Treasury bonds by foreign debt holders, the use of deepfakes to disseminate misinformation, and “meme engineering” to manipulate beliefs and behaviors. While these threats pose a limited risk to the economic situation in the short term, they may worsen over time due to the gradual erosion of resilience and financial capacity to respond to these developments.
In this context, the research by Tobias Sytsma and others, titled “Technological and Economic Threats to the U.S. Financial System: An Initial Assessment of Growing Risks,” published by the RAND Corporation in 2024, is significant. The study explores emerging patterns of threats to the financial system, particularly those arising from social media and advancements in artificial intelligence, as well as the determinants of these threats’ impacts and the requirements for addressing them.
Diverse Threats
There are four primary patterns of threats facing the U.S. financial system:
Cyber Attacks and Data Poisoning: These are among the most widespread threats due to the increasing use of AI and machine learning to manipulate data, including hacking, espionage, and asset theft. In the U.S. trading market, 80% of trading volume is executed by AI algorithms that allow for rapid information processing and decision-making faster than humans. Despite this importance, these algorithms present challenges in light of the growing number of hostile attacks and data poisoning attempts. Hostile attacks significantly affect trading operations by deliberately manipulating and introducing disruptions in trading data, while data poisoning attempts to insert incorrect patterns and biased forecasts, leading to inaccurate predictions regarding trading outcomes. This pattern carries substantial risks for the financial system, as defending against hostile attacks is challenging due to the myriad ways an adversary can manipulate the inputs fed into the model. Moreover, hostile attacks on deep learning trading models have led to significant financial losses ranging from 23% to 32% in trading portfolios as they result in suboptimal trading decisions and inaccurate forecasts about potentially profitable investments.
Bond Flooding: This threat aims to weaken the U.S. economy by intentionally selling large quantities of U.S. Treasury bonds to create chaos in the market. This pattern has a significant impact as markets and governments cannot plan for it. The flooding process involves a one-time massive sell-off, causing bond prices to drop and interest rates to rise, resulting in widespread chaos shaking the markets. This threat pattern primarily originates from China, which holds more U.S. Treasury bonds than any foreign country besides Japan, coupled with the changing geopolitical relationship between the U.S. and China. This tactic was notably employed during the 2008 financial crisis when Russia urged China to work together to offload U.S. government debt to disrupt the American market. It resurfaced in 2022 and 2023, as China continued to reduce its holdings of U.S. Treasury bonds. However, U.S. financial markets were able to absorb this shock, thanks to the inherent resilience of the American market due to its size and available liquidity, along with intervention from the Federal Reserve. For example, in 2008, China and Russia sold $220 billion in U.S. government securities, prompting a rise in interest rates by about 1.4 percentage points. Nevertheless, this strategy did not adversely affect the U.S. market, thanks to prompt governmental responses aimed at mitigating fallout, including relatively swift tightening of monetary policy by the Federal Reserve to purchase large volumes of assets, stabilizing interest rates and reducing volatility in the short term. The Federal Reserve subsequently raised five-year Treasury bond rates by about 4 percentage points. While bond flooding may cause short-term fluctuations and significant losses for some bondholders, it is unlikely to destroy the economy. Although China’s selling actions in 2008 were an aggressive financial policy aimed at attaining political goals, they ultimately failed, and since then, China has been unable to influence U.S. monetary policy, reflecting the reality that China’s market power is insufficient to generate meaningful geopolitical concessions.
Deepfake Information: One of the most dangerous threats to the U.S. financial system is the use of advanced technology to spread fake materials and misinformation. Deepfake attacks involve altering or fabricating videos, audio recordings, images, or text to create false impressions, disseminated via social media to influence or manipulate public opinion, targeting consumers, investors, businesses, and policymakers. Typically, deepfakes are created using rapidly evolving AI tools, and companies targeted by deepfakes face damage to their reputation, sales, and stock prices. Fake images and videos pose a significant risk to financial markets. A notable example occurred on May 22, 2023, when a fake image of an explosion outside the Pentagon began circulating on social media, and although it was later revealed to be false, it caused the S&P 500 index to drop by 0.3 percentage points within minutes. Additionally, voice cloning to target individuals has significant implications for the financial system. In 2019, scammers managed to steal 667millionfromU.S.residents,andAmericancompanieslostover667millionfromU.S.residents,andAmericancompanieslostover1.7 billion due to voice cloning fraud in the same year. Two factors influence the magnitude of the impact from deepfake operations: firstly, the adversary’s capability to engage in a deepfake campaign spanning various media forms; secondly, the nature of the threat itself. The financial system may be more vulnerable to damage if an adversary releases a fake video of the President of the United States or the Secretary of Defense, potentially leading to political chaos that impacts financial markets.
Meme Engineering: This threat pattern is based on the strategic manipulation of ideas, concepts, or beliefs to influence a targeted audience, leveraging psychological triggers or cognitive biases to change attitudes, beliefs, and behaviors. Meme engineering involves designing viral ideas or culture to achieve specific psychological or material outcomes. The term is derived from genetic engineering, as memes are the cultural equivalent of biological genes; more specifically, memes are multifaceted cultural units, such as images, videos, phrases, or concepts, that spread rapidly online through imitation and sharing. Historically, Russia has been the primary actor in using covert meme engineering. However, China has begun shifting its focus from large-scale meme engineering operations on social media to those targeting specific individuals. Meme engineering relies on two patterns: the first is a rapid attack aiming to quickly degrade a specific set of stocks; the second is a long-term deterioration of beliefs, ideologies, or trust in the market or financial institutions. A common example is the involvement of a Russian internet research agency in the 2016 U.S. presidential elections by strategically spreading false social media content designed to maximize manipulation of American voters’ beliefs and actions. Meme engineering attacks that promote long-term deterioration of beliefs and values present a significant challenge and depend on the ability to create credible misleading information that can be widely disseminated by leveraging AI, increasing the likelihood of a successful meme engineering attack.
Multiple Determinants
The patterns of threats to the U.S. financial system entail specific direct costs for markets and indirect costs affecting norms or institutions. Several factors influence the deployment of these threat patterns and their repercussions:
Technological Capabilities: These capabilities contribute to the likelihood of a successful attack through technological enhancements or increased technology integration. For instance, deepfake attacks become more likely with improvements in AI technologies, making false information appear more realistic and harder to detect.
Contextual Incentives: These largely depend on the idea of existing context and circumstances, such as market uncertainty, volatility, or public relations crises. Geopolitical tensions also serve as significant incentives that may alter the motivations of foreign actors by reducing the political costs associated with economic strikes, particularly since attacks aimed at destabilizing the U.S. financial sector may have backlash effects on the economy of the attacking country, thereby diminishing incentives. However, some countries outside of the U.S. financial system, like North Korea or those under economic sanctions like Russia, may have the incentive to harm the American economy.
Risk Management: Risk management involves the various measures taken to reduce the cost of an attack on the financial system. These include complementary measures to disrupt financial AI systems and coordinated data governance policies.
In conclusion, the U.S. financial system is a crucial pillar in managing risks for the global market. This significance makes the system a vital target for countries in conflict with the United States, employing patterns and threats that impact financial stability within the nation. However, most threats identified may cause more limited economic losses if they occur in isolation; yet, if the U.S. financial system faces a series of incremental, interconnected risks, it may lead to widespread effects.
Addressing these threats requires adopting policies focused on the responsible development and application of AI technologies, serving as important safeguards for financial stability while imposing export controls on advanced semiconductors and obliging cloud service providers to inform the U.S. government as part of regulatory policies. Additionally, economic policies should be formulated to enhance the resilience of the financial sector by encouraging competition in the market, leading to a broader range of options available for financial institutions looking to integrate AI into their operations.
Finally, there is a need to engage stakeholders to achieve a more advanced understanding of how the threats to the U.S. financial system are evolving, especially over the next decade. Continuous monitoring and adaptation to the changing landscape, responsibly embracing technological advancements, and enhancing competition and transparency policies, will enable the United States to mitigate these emerging threats and maintain its financial stability.
Source:
Sytsma, T., Marrone, J. V., Shenk, A., Leonard, G., Grek, L., & Steier, J. (2024, July). Technological and economic threats to the U.S. financial system: An initial assessment of growing risks. RAND Corporation. https://www.rand.org/pubs/research_reports/RRA2533-1-v2.html
