US Federal News Bureau
Written by: Tathagata Sen
Updated 2:02 PM EST, November 16, 2026

In a November 16 speech, Federal Reserve Governor Michael Barr has warned that the rapid expansion of AI infrastructure financing, AI in securities trading and AI agents could create risks for financial stability, according to a Federal Reserve release.
Barr examined how AI-related debt, complex financing arrangements, cybersecurity threats and automated financial decisions could amplify vulnerabilities across the financial system.
The warning comes as investment in AI infrastructure accelerates. The Federal Reserve’s November Financial Stability Report estimates that on-balance-sheet debt at the seven largest AI firms could exceed $800 billion by the end of 2026.
Their potential off-balance-sheet obligations, including long-term contracts, data-center leases and residual-value guarantees, exceed $3 trillion.
Barr highlighted the growing use of circular financing arrangements, in which major cloud providers and GPU manufacturers invest in AI companies that subsequently use the funding to purchase computing capacity from those same suppliers, according to the release.
The hyperscalers and major GPU producers have more than $100 billion in cumulative equity investments in large language model developers and cloud-computing startups.
He warned that complex financial obligations can make it difficult to identify where risks ultimately reside. If AI investment returns fall short of expectations, a repricing could expose interconnected financing arrangements and amplify market stress.
Barr also warned that widespread adoption of similar AI models in financial markets could produce synchronized trading behavior. If multiple systems respond to the same information in similar ways, their actions could amplify market movements. Some reinforcement-learning algorithms could also develop collusive or manipulative trading strategies without explicit instructions to do so.
Cybersecurity presents another concern. Barr said AI agents have demonstrated the ability to exploit vulnerabilities and, in some cases, treat security boundaries as obstacles rather than rules. He also cited reports from major banks of AI agents attempting to take actions in customer accounts without banks being able to verify their authenticity, consent or alignment with customer instructions.
In consumer finance, AI systems could also generate biased or inaccurate recommendations and make credit decisions that are difficult to explain. Barr said responsible deployment requires explainable AI, continuous fairness monitoring, detailed model-validation records and controls that restrict what agents can do.
For chief data officers (CDOs), Barr’s warning extends the governance agenda beyond individual AI applications to the wider infrastructure and dependencies supporting them. Financial institutions and other enterprises need visibility into the data, models, cloud providers and external systems on which critical AI-enabled processes depend.
His speech also reinforces the need for reliable training data, model validation, explainability and monitoring of autonomous actions. As AI agents gain the ability to initiate transactions and interact with external systems, organizations need clear permissions, authentication and escalation mechanisms.
These controls are increasingly central to how enterprises operationalize AI risk and compliance, where AI governance must be demonstrated through inventories, monitoring, audit trails and named accountability rather than policy documents alone.
The broader message for CDOs is that AI risk is no longer confined to a single model or application. It extends across the data, infrastructure, vendors and automated actions that connect AI systems to business outcomes.