🔍 Read the full analysis: The Fed Survey’s AI Risk Figure Is 87%; Australia Plans To Put AI To The Test on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
In its November Financial Stability Report, the Federal Reserve said 87% of 23 market contacts surveyed in September and October named AI as a risk to U.S. financial stability, up from 50% in the spring. Australia’s APRA also warned about frontier AI-related cyber threats and said its 2027 system-wide stress test will examine AI-driven disruption.
The Federal Reserve’s November Financial Stability Report says 87% of 23 market contacts surveyed in September and October named artificial intelligence as a risk to U.S. financial stability, up from 50% in the spring. Separately, Australia’s prudential regulator, APRA, said it will examine AI-driven disruption in a 2027 system-wide stress test and warned that advanced models are accelerating cyber threats.
The Fed’s survey was conducted by staff at the Federal Reserve Bank of New York. Respondents included professionals from broker-dealers, banks, investment funds and advisory firms. The report says the survey summarizes those contacts’ views, not the views of the Federal Reserve or the New York Fed. Its data run through October 28, so the survey responses predate developments in November.
AI was the most-cited risk in the survey. Other concerns included a correction in risk assets, named by 52% of respondents; higher long-term interest rates, 48%; cyberattacks, 43%; and fiscal debt sustainability, also 43%. The Fed said mentions of AI, higher long-term rates and cyberattacks rose most from the spring survey. Contacts rated cyberattacks and fiscal debt sustainability as having the most severe potential consequences.
Respondents described several different AI-related risks. They pointed to the possibility that AI-related stocks could fall if returns fail to justify large investments, and cited stretched valuations, debt and complex financing as potential amplifiers. They also warned that AI could lower the barrier to sophisticated cyberattacks, including unintended damage caused by frontier models. The survey further recorded concern about complex AI-linked financing arrangements.
The report also sets out the scale of borrowing connected to advanced computing. It says related debt issuance exceeded $680 billion by October 2026, about 31% of investment-grade nonfinancial corporate issuance, compared with less than 6% in 2023. The largest U.S. bank holding companies had close to $1 trillion in exposure to firms and vehicles behind the buildout, according to the report.
APRA’s System Risk Outlook, published November 12, said frontier models, including those from Anthropic and OpenAI, are improving at finding and exploiting software flaws. The Australian regulator also identified reliance on shared AI providers as a potential source of system-wide vulnerability: an outage or failure at a provider used by many institutions could affect them at once. APRA found that few institutions had tested how they would switch away from critical AI providers.
Explained in motion
Banks, insurers and funds often rely on the same few models, platforms and service providers. Supervisors warn that one problem there can reach all of them at once.
Illustrative sequence, not a real incident
Shared providerMany institutions run on the same model, platform or service provider.
One changeA failure, a compromise or a model update changes how that provider’s AI works.
Many hit at onceEvery institution that depends on it is affected at the same time.
Services stallPayments, claims or trading slow down. The Fed warns that disruptions can hurt market liquidity and confidence.
No quick switchThe damage grows when there is no ready substitute for the service.
Source: APRA, System Risk Outlook (November 12, 2026) and letter to industry on AI (April 30, 2026); Federal Reserve, Financial Stability Report (November 2026), pp. 47–48. Illustrative sequence built from the mechanisms they describe.
Graphic: Powered by Thorsten Meyer AI
AI Risks Reach Banks and Markets
The two regulators are addressing different parts of the same exposure. The Fed’s survey captures concern among market participants about asset prices, cybersecurity and financing, while its report also describes potential consequences if attacks or service disruptions affect financial institutions and their providers. APRA is preparing to test how AI-related shocks could spread through Australia’s financial system.
Shared providers matter because institutions may depend on the same external services. The Fed warns that disruption at a critical third-party provider could affect multiple institutions, especially where substitutes are limited. In its account, a disruption at a financial institution could contribute to weaker market liquidity, loss of depositor or investor confidence, and forced asset sales. These are risks described by the report, not outcomes it says have occurred.
The borrowing figures also show that AI infrastructure is linked to a sizeable and growing financing effort. If expected returns disappoint, the Fed’s report says, valuations and debt-backed investment could come under pressure. The report does not predict that a correction will happen; it records the concern and provides figures on the financing involved.
Products Worth Considering
Dell UltraSharp U2724DE 27-Inch WQHD IPS Black Monitor
Canon SELPHY CP1500 Compact Photo Printer Bundle with Ink and Paper Set, White
Vari Ergo Electric Standing Desk, 54×26, Light Wood/White
As an affiliate, we earn on qualifying purchases.
Two Supervisors, Different Evidence
The Fed’s survey is a twice-yearly exercise asking market contacts which shocks could most damage the U.S. financial system over the next 12 to 18 months. This edition collected responses from 23 contacts in September and October, compared with 20 contacts in March and April. The difference between the reported shares is a change in survey responses, not a measure of the probability that AI will cause a crisis.
The report distinguishes the contacts’ concerns from the Board’s own assessment. It describes the banking system as “sound and resilient”, while noting high equity valuations partly associated with enthusiasm for superintelligence. It also says advances in large language models and agentic systems have created new challenges for protecting financial institutions, market infrastructure and third-party service providers. The report says accounts of agentic systems being involved in cyber events increased since May, but gives no count.
APRA’s outlook adds a supervisory response focused on testing. Its planned 2027 exercise will look at AI-driven economic disruption, alongside the regulator’s concerns about cyber capabilities and dependence on common providers. The source material does not describe the exercise’s full scenarios or methodology.
“Frontier AI models, including those from Anthropic and OpenAI, are speeding up cyber threats.”
— APRA System Risk Outlook
Limits of the Survey and Test
The Fed’s 87% figure comes from a small panel of 23 contacts, whose identities are not disclosed in the supplied material. It records what those respondents cited as a risk; it does not establish the likelihood or scale of a future AI-related shock. The report also stresses that these are respondents’ views rather than the Fed’s own.
It is not clear from the available information how APRA will design its 2027 scenarios, which institutions or AI services will be covered, or how it will measure readiness to switch providers. The Fed gives no count for reported cyber events involving agentic systems. Nor does the material establish that the AI-related borrowing or valuations will lead to losses.
Stress Testing Moves to 2027
APRA’s next stated milestone is its 2027 system-wide stress test, which will examine AI-driven economic disruption. Further details about its scope and scenarios have not been provided in the source material. The Fed’s report presents the survey and its own risk assessment; no new U.S. stress-test timetable is specified there.
For now, the central questions are whether financial institutions can withstand AI-enabled cyber incidents, how exposed they are to common technology providers, and whether financing for AI infrastructure is sustainable if expected returns fall short. The reports identify those concerns, but do not determine how likely any one outcome is.
Key Questions
What does the Fed’s 87% figure measure?
It is the share of 23 market contacts surveyed in September and October who named AI as a risk to U.S. financial stability. It is not a forecast of an 87% chance of an AI-related crisis.
Does the Fed say AI will cause a financial crisis?
No. The report records concerns raised by survey respondents and discusses possible risks, including cyberattacks, valuation pressure and financing. It does not say that a crisis will occur.
What is Australia planning to test?
APRA says its 2027 system-wide stress test will examine AI-driven economic disruption. Details of the test’s scenarios and participating institutions were not provided in the source material.
Why are shared AI providers a concern?
If many financial institutions rely on the same critical provider, an outage or other disruption could affect multiple firms at once. The Fed warns that limited substitutes could amplify the effects; APRA says few institutions had tested a way to switch providers.
How much AI-related borrowing did the Fed report?
The report says debt issuance tied to advanced computing exceeded $680 billion by October 2026. It represented about 31% of investment-grade nonfinancial corporate issuance, compared with less than 6% in 2023.
Source: ThorstenMeyerAI.com
