Artificial intelligence has emerged as the biggest threat to the US financial system, according to a Federal Reserve survey released Friday, with concerns about the technology vastly outweighing other risks to stability. The central bank's twice-yearly assessment found that 87% of respondents cited AI as a salient risk to the system, up sharply from 50% in the spring survey earlier this year. This far exceeded mentions of the next most-cited danger, a risk asset correction, which was flagged by 52% of those surveyed.
The Fed report identified three specific areas where AI poses financial stability concerns. First, respondents pointed to elevated equity valuations and the possibility that lower-than-expected earnings could trigger a market correction. Second, they highlighted cybersecurity risks stemming from either malicious attacks by state actors or damage caused by frontier AI models themselves. Third, they noted risks tied to complex forms of AI-related financing arrangements.
Other leading risks mentioned by survey respondents included higher long-term interest rates (cited by 48%), cyberattacks (43%) and fiscal debt sustainability (43%), the latter reflecting concerns about the US debt exceeding $40 trillion and the budget deficit reaching nearly $2 trillion. The respondents surveyed included a variety of market participants, economists, analysts and institutional sources, with the report covering the period through October 28. The survey comes as Wall Street watches AI companies closely and communities grapple with rapid expansion of data centers powering the technology, with some 56% of voters saying data centers were of crucial or major importance in their choices during this month's midterm elections.
Beyond AI risks, the Fed's assessment otherwise found the financial system fairly stable. However, the report noted that hedge fund leverage stands at its highest level since 2013 and remains concentrated among a relatively small group of funds, presenting a potential vulnerability should market conditions shift.