TL;DR
In its November 2026 Financial Stability Report, the Federal Reserve said 87% of 23 surveyed market contacts cited AI as a risk to U.S. financial stability, up from 50% in spring. Separately, Australia’s APRA warned about cyber threats from frontier AI models and said its 2027 system stress test will examine AI-driven disruption. The findings point to risks for Australian institutions, but they do not establish that a financial crisis or AI-related failure is underway.
The Federal Reserve reported on November 13 that 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. A day earlier, Australia’s prudential regulator, APRA, warned that frontier AI models could intensify cyber threats and said its 2027 system-wide stress test will examine disruption driven by AI, putting the issue on the agenda for Australian banks and other financial institutions.
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
The New York Fed gathered views from professionals at broker-dealers, banks, investment funds and advisory firms about shocks that could harm the U.S. financial system over the next 12 to 18 months. The 87% figure reflects 23 respondents; the spring comparison was 50% of 20 contacts surveyed in March and April. The report says the survey summarizes respondents’ views and does not represent the views of the Federal Reserve Board or the New York Fed. It is a small panel, not a forecast or a measurement of the likelihood that a crisis will occur.
Respondents identified several different exposures under the AI label. They cited the possibility of a sharp correction in AI-related shares if large investments fail to produce expected returns, with high valuations, debt and circular financing potentially adding to the risk. They also pointed to AI-enabled cyberattacks and possible unintended harm from frontier models, as well as complex financing linked to AI. The Fed separately reported that debt issuance tied to advanced computing had exceeded $680 billion by October 2026, about 31% of investment-grade nonfinancial corporate issuance, compared with less than 6% in 2023. It also said the largest U.S. bank holding companies had close to $1 trillion of exposure to firms and vehicles behind the buildout.
APRA’s November 12 System Risk Outlook said frontier models, including those from Anthropic and OpenAI, are improving at finding and exploiting software flaws. The regulator’s concern is not limited to attacks: financial firms’ dependence on a small number of AI providers could spread an outage, compromise or disruptive change across multiple institutions. APRA said few institutions in its review of large banks, insurers and pension funds had tested how to switch away from critical AI providers.
For Australian banks, the findings highlight a risk that can cross institutional boundaries. If several firms rely on the same AI model, cloud platform or related service, a failure or security incident at that provider could affect many of them at once. Depending on how the service is used, disruption could slow payments, insurance claims or trading, and make it harder for firms to maintain service while they recover. APRA’s review suggests that tested alternatives may be lacking at some institutions, although its published summary does not quantify the extent of that weakness across the entire sector.
The Fed’s survey also links AI to financial stability through investment and financing, not just technology operations. If expectations for AI earnings fall short, a correction could affect companies and lenders exposed to the buildout. The figures on U.S. issuance and bank exposures describe the U.S. system; they should not be treated as measurements of Australian banks’ exposure. For Australian readers, the immediate relevance is APRA’s stated attention to cyber risk, provider concentration and system-wide disruption.
Products Worth Considering
Samsung 32-inch Odyssey G5 G51F QHD 180Hz Gaming Monitor
Amazon Kindle Scribe Colorsoft 64GB with 11-Inch Color Display and Premium Pen, Graphite
Apple Watch Series 12 GPS 42mm
As an affiliate, we earn on qualifying purchases.
What the Two Supervisors Measured
The Fed’s Financial Stability Report uses a twice-yearly survey to identify risks market contacts believe could damage the U.S. financial system over the following 12 to 18 months. The November report drew on responses collected in September and October and says its underlying data run through October 28. Its AI result is a count of views in that survey, rather than a formal assessment that AI itself has caused financial instability.
APRA’s System Risk Outlook is a separate Australian supervisory assessment. It described how frontier AI could change the cyber threat environment and identified possible dependence on shared providers as a system risk. The two publications do not establish a direct link between the U.S. survey result and Australian bank losses. They do show that supervisors in both countries are examining how AI investment, cyber capabilities and concentrated service dependencies could affect financial firms.
“Frontier models, “including those from Anthropic and OpenAI”, are speeding up cyber threats.”
— APRA, System Risk Outlook
Exposure and Readiness Remain Unclear
The survey does not show how likely any of the cited risks are, or how much damage a particular AI-related shock would cause. Its 23 U.S. respondents form a small panel, and their answers are not a census of financial institutions. The report’s exposure figures also concern large U.S. bank holding companies and cannot establish the scale of Australian banks’ direct or indirect exposure.
APRA’s published findings do not specify which institutions had tested provider switching, how many had not, or whether any shared-provider incident has disrupted Australian financial services. It is also unclear how the 2027 stress test will define AI-driven economic disruption, which institutions and scenarios it will cover, and what results APRA will publish. The material describes risks and planned supervisory work, not a confirmed AI-caused failure.
APRA’s 2027 Stress Test
APRA says its next system-wide stress test, planned for 2027, will examine AI-driven economic disruption. That exercise should give the regulator a way to assess how financial institutions and the wider system might respond to relevant shocks, though the source material does not specify the scenarios or timetable beyond the year.
In the meantime, the key developments to watch are further detail from APRA on the test and any additional supervisory guidance on cyber resilience and dependence on critical AI providers. Banks, insurers and pension funds will also face questions about whether they can continue essential services if an AI supplier becomes unavailable or compromised. The available reports do not say what measures individual Australian institutions have taken.
Key Questions
What does the Fed’s 87% AI risk figure mean?
It means 20 of the 23 market contacts surveyed in September and October named AI as a U.S. financial stability risk. It records respondents’ views, not the Fed’s estimate of the probability of a crisis.
Does the figure describe Australian banks?
No. The survey concerns U.S. market contacts. APRA’s separate findings address risks to Australian institutions, but the supplied material does not quantify Australian banks’ AI-related financial exposures.
What AI risks did the reports identify?
The Fed’s respondents cited potentially overstretched AI-related valuations, cyber threats and complex AI financing. APRA highlighted frontier models’ ability to find and exploit software flaws, as well as the risk that dependence on shared providers could spread disruption.
What is APRA planning to do?
APRA says its 2027 system-wide stress test will examine AI-driven economic disruption. The published material does not yet describe the test scenarios or its detailed timetable.
Source: Thorsten Meyer AI
