Top federal regulators have formally recognized artificial intelligence (AI) as a potential risk to the financial system. The Financial Stability Oversight Council has classified AI as an emerging vulnerability, highlighting the need for its careful implementation and supervision to handle potential risks.
AI has vast capabilities, such as the ability to generate song lyrics, write stories, or create images instantly. Its strength, however, brings certain dangers. Increasing concern is being raised by U.S. officials about the potential risks AI poses in real-world scenarios, notably financial markets.
One such concern is potential bias in AI decision-making. In financial situations, AI bias could result in individuals being denied mortgages or car loans based on factors such as skin color or gender, introducing unfair prejudice into financial decision-making processes.
Furthermore, the threat of cybersecurity breaches is significant. There is a risk that AI tools used by banks could be exploited by hackers, potentially granting unauthorized access to the financial system. This could result in customers being unable to access their bank accounts, disrupting crucial financial operations.
Another issue related to the use of AI technology is its tendency to create 'hallucinations', or output presented as factual despite having no grounding in reality. For instance, a recent lawsuit in New York cited four cases procured by an AI tool, ChatGPT, which turned out to be non-existent, demonstrating the potential risk to accuracy and trust in financial scenarios.
Regulators are encouraging the financial industry to remain vigilant, indicating they will be closely inspecting how AI models operate. Banks are being cautioned to avoid using flawed models and, in essence, to self-police their use of AI. However, striking a balance between appropriate regulation and fostering innovation is crucial. Too much regulation may stifle the growth and development of AI, while insufficient regulation may expose the financial system to the risks discussed above. Therefore, a middle ground where innovation is encouraged while potential risks are managed effectively is essential.
