Fed Survey Contacts Name AI as Top Stability Risk

Contacts in a New York Fed survey have named artificial intelligence as the most-cited near-term risk to US financial stability. The November 2026 Financial Stability Report, which generally reflects information as of 28 October, includes a survey box in which 87% of respondents cited AI, up from 50% in the spring. A correction in risk assets was named by 52%, higher long-term rates by 48%, and cyberattacks and fiscal debt sustainability by 43% each.

Those figures come from a staff outreach exercise, not from a Board forecast. The report was posted by @federalreserve at 21:18 GMT on 13 November with the Board’s current assessment of the stability of the US financial system. The survey box itself is narrower. It says it is a summary of respondents’ views and “should not be interpreted as representing the views of the Federal Reserve board or the Federal Reserve bank of New York.”

The Fed survey of 23 contacts put AI first, and split that concern into three parts

Box 5.1, “Survey of Salient Risks to Financial Stability,” says staff from the Federal Reserve Bank of New York received responses from 23 contacts during September and October, including professionals at broker-dealers, banks, investment funds and advisory firms. The spring comparison chart is sourced to a survey of 20 market contacts from March through April. The samples are small, and the report does not present them as a census of the market.

“AI was the top-cited risk, with responses noting concerns around elevated equity valuations, AI-enabled cyberattacks, and complex forms of AI-related financing,” the box says. It then sets out “three distinct areas of concern: (1) elevated equity valuations and the possibility that lower-than-expected earnings could result in lower valuations and a market correction; (2) cybersecurity risk, stemming from either a malicious attack by sophisticated state actors or damage caused by frontier AI models themselves; and (3) risks associated with complex forms of AI-related financing.”

Frequency is not the same as severity in this survey. “Cyberattack and fiscal debt sustainability risks were viewed as having the most severe potential consequences,” even though both were cited by a smaller share of contacts than AI. The overview section of the report lists the same five headings as the risks most frequently named by participants, and points readers to the survey box rather than restating them as the Board’s own ranking.

That mix of valuation, cyber and financing worries has already appeared in other official financial-stability writing this autumn. The IMF’s October Global Financial Stability Report named an AI market correction, rising bond yields and cyberattacks as near-term risks. The Financial Stability Board has also finalised 12 sound practices for how banks and other financial firms adopt AI. Those are separate institutions and separate documents. They show that the New York Fed contacts are not raising an isolated set of themes.

A separate report box puts advanced-computing debt above $680bn

The survey’s third area of concern, complex AI-related financing, sits beside a different part of the same report. Box 2.2, “Leverage in the Data Center Buildout,” is a data discussion, not a poll. It says: “Advanced computing–related debt issuance rose to more than $680 billion by October 2026, about a fivefold increase from $130 billion in all of 2023.” The same box says advanced computing–related borrowing as a share of nonfinancial investment-grade corporate debt issuance “increased from less than 6 percent in 2023 to about 31 percent in 2026.” Outstanding public debt related to advanced computing “is estimated at more than $1.1 trillion as of October 2026.”

The box also warns that sizable transactions occur in private credit, “making the exact magnitude of advanced computing–related debt outstanding more difficult to measure.” That is a measurement caveat, not a forecast of default. It does, however, give a concrete official backdrop to the contacts’ talk of complex financing, including the lease commitments and backstop guarantees the box says large technology firms are increasing. TSN has already reported one such structure, Google’s backstop of financing for TeraWulf’s $19bn Anthropic leases. That deal is not named in the Fed report. It is an example of the kind of contingent support the box describes.

Political debate about who would bear losses in a sharper AI downturn is already under way in Washington. House Democrats led by Representatives Casar and Waters have argued against any AI-bubble bailout. The Fed report does not address a bailout, and the survey should not be read as a comment on that campaign.

A 23-contact survey is not a Board verdict on AI risk

The most important line in Box 5.1 is the one that limits what the figures can be used to say. Twenty-three contacts, reached by New York Fed staff, named AI more often than any other heading. That is a clear survey result. It is not the Federal Reserve Board declaring AI to be the biggest risk to the financial system. The box says so directly, and TSN should keep that distinction.

Even inside the survey, the ranking is not a single number. AI leads on how often it was mentioned. Cyberattacks and fiscal debt sustainability lead on how severe respondents thought the consequences could be. A reader who stops at “87%” will miss that split, and will also miss how small the panel is. A shift from 50% to 87% is a large move in a 20-to-23-person sample, and it is still a move in a sample.

The debt figures in Box 2.2 should be kept in their own box as well. They are official estimates of issuance and of outstanding public debt, with a private-credit gap acknowledged, as of October 2026. They help explain why financing methods are on the contacts’ list. On their own, those figures say nothing about whether those debts will fail, or that the Board has judged the buildout unsound.

The next official facts worth waiting for are ordinary ones: how the same contacts answer in the next survey, and whether later Fed data tighten the measure of private advanced-computing debt that Box 2.2 says is hard to capture. Until then, the story is a small market survey with AI at the top of the citation list, a separate official box showing a sharp rise in related borrowing, and a report that asks not to have the first of those read as the second.