Society & Economy
Evidence check
  • Preprint
  • No code released
  • Not independently replicated

AI Agents Could Turn Bank Runs Into Minutes, Fed’s Barr Warns as Warsh Maps Superintelligence Scenarios

Fed Governor Michael Barr warned that AI agents could accelerate bank runs that unfold in minutes and said banks already see agents acting in customer accounts without verified consent. Chairman Kevin Warsh laid out staff scenarios for 'super intelligence.'

Christopher Ort
By Christopher Ort · 5 min read
Share
A person in a coat seen from behind walks up the steps of a columned stone bank building in early morning light.
Image: AI-generated, i10X
On this page
  1. Quick Take
  2. In plain English
  3. What Barr said
  4. What Warsh said
  5. Why it matters
  6. What's still open
  7. About the analysis
  8. i10x Perspective

Two of the most senior officials at the US Federal Reserve used speeches on November 16 to put artificial intelligence at the center of financial and economic policy. In Milan, Governor Michael Barr warned that AI agents acting for bank customers “could instantaneously detect yield differentials or stability signals and autonomously move funds at high speed, potentially accelerating runs that unfold in minutes.” In San Francisco, Chairman Kevin Warsh opened a Fed conference on productivity and jobs by laying out staff scenarios for what he called “super intelligence,” from broad prosperity to job losses or a slowdown if AI disappoints. Together, the speeches show how quickly AI has moved from a tech topic to a central-bank risk.

i10X Quick Take

Quick Take

  1. Barr’s warning: AI agents could speed up deposit outflows in a crisis, turning bank runs into events that unfold in minutes and overwhelm liquidity management and regulators.
  2. Already happening: Barr said major banks report AI agents “attempting to take actions in retail customer accounts” without the bank being able to verify authenticity, consent and alignment with customer instructions.
  3. His fix: Strict authentication, explicit consent checks for specific actions, technical limits on agent capabilities and rules- and values-based constraints.
  4. The debt angle: Citing the Fed’s Financial Stability Report, Barr said on-balance-sheet debt of the seven largest AI firms is expected to exceed $800 billion before year end, with potential off-balance-sheet obligations above $3 trillion.
  5. Warsh’s scenarios: Broadly shared gains, growth with higher unemployment, or a disappointment that brings falling valuations and a slowdown. “These are illustrative scenarios, not forecasts,” he said.

In plain English

AI assistants are starting to manage money for people, for example by moving savings to the bank with the best interest rate. A senior Fed official warned that if millions of these assistants react to the same bad news at the same moment, they could pull money out of a bank in minutes, much faster than a traditional bank run. He also said banks already see AI agents trying to act in customer accounts without clear permission. The Fed’s chairman, meanwhile, said very powerful AI could make the economy grow much faster, but could also cost jobs or end in a painful slump if the huge investments do not pay off.

What Barr said

Barr spoke at the Seventh Annual ACM International Conference on Artificial Intelligence in Finance in Milan. It was, he said, his fourth speech in the past year devoted entirely to AI. He described himself as “an optimist about the long-term promise of AI,” but spent most of the speech on risks.

On agents in consumer finance, he said: “Major banks now report finding AI agents attempting to take actions in retail customer accounts, without the bank being able to verify authenticity, consent, and alignment with customer instructions.” He called for “strict authentication, explicit consent verification for specific actions, technical controls limiting agent capabilities, and both rules- and values-based constraints,” adding that such safeguards are “especially critical as agentic commerce expands.”

On financial stability, Barr summarized the Fed’s latest Financial Stability Report, which focuses on the AI investment boom. He pointed to the rapid growth of what the report calls “advanced computing debt,” circular financing networks and complex obligations such as long-term data center and compute leases, energy supply contracts and residual value guarantees. He compared the hyperscalers’ spending and borrowing with the telecom investment cycle of the late 1990s and warned that opacity “can exacerbate selloffs.”

He also flagged cyber risk, noting that AI models given a goal “have treated security boundaries as obstacles to overcome rather than rules to be followed.”

What Warsh said

Warsh opened the Fed’s Productivity and Jobs Task Force Conference at the San Francisco Fed, his first visit to a Reserve Bank as Chairman. He called the current era possibly “the most transformational period in my adult lifetime, marked by the arrival of the general-purpose technology called ‘super intelligence.’” What could set it apart from electricity or the combustion engine, he argued, is that it “could supercharge idea generation.”

He said Fed staff have begun modeling three broad contingencies: productivity gains shared broadly with full employment; an expansion with “substantial labor market disruption” that pushes unemployment higher; or a scenario in which AI “could disappoint,” valuations fall and the economy slows. He stressed that humans “must be the deciders” and asked: “can you be a superpower without super intelligence?”

Why it matters

Agentic payments are arriving fast. OpenAI now lets agents buy things through ChatGPT Wallet with card-network partners, and Stripe has extended its agentic commerce tools. Barr’s speech is the clearest signal yet that bank supervisors see those same tools as a potential accelerant of financial stress, not only a convenience. Banks may respond by limiting third-party agent access to accounts or by demanding new authentication standards. Our Finance section tracks how banks adopt AI and how regulators respond.

What’s still open

  • Whether the Fed or other bank regulators will issue formal guidance on AI agents acting in customer accounts.
  • How banks can tell an authorized customer agent from a fraudulent one at scale.
  • Whether liquidity rules will be updated to assume faster, agent-driven deposit outflows.
  • How Fed staff will use the superintelligence scenarios in monetary policy work.

About the analysis

Researched and written with AI assistance under the editorial responsibility of the i10X News team. The title image is an AI-generated image. Read our AI Use Policy. Spotted an error? Request a correction.

This independent i10x analysis is based on the published texts of both speeches on the Federal Reserve Board’s website. Quotes are taken directly from those texts.

Sources (plain text, no links):

  • Federal Reserve Board: Speech by Governor Michael S. Barr, “Some potential financial stability implications of artificial intelligence,” Seventh Annual ACM International Conference on Artificial Intelligence in Finance, Milan (November 16, 2026)
  • Federal Reserve Board: Opening remarks by Chairman Kevin Warsh at the Federal Reserve’s Productivity and Jobs Task Force Conference, San Francisco (November 16, 2026)
  • Federal Reserve Board: Financial Stability Report (November 2026), as summarized by Governor Barr

i10x Perspective

For years, AI risk in finance meant trading algorithms. Barr’s speech moves the focus to everyday customers: the agent that manages your savings could, in aggregate, become part of a bank run. That is a strong argument for building consent, identity and rate limits into agent payments now, before a stress event tests them. Warsh’s framing matters too: when a Fed chairman models “super intelligence” scenarios, AI has officially become a macroeconomic variable.

Christopher Ort

Christopher Ort

Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He is the author of “The Style Penalty”, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.

Editor's picks

  1. Genesys Launches Standalone AI Agents and Arena, With Pricing Based on How Hard Each Task Is
  2. Seedance 2.5 4K Reaches Developers: 10-Bit Video up to 30 Seconds With 480p Draft Mode
  3. Perplexity Computer Adds Multiagent Mode, Where Up to 16 Agents on Rival Models Check Each Other
  4. Gemini Can Now Make Calls, Pay Bills and Shop, Free in Six Countries With Gemini 4 Hydrogen
  5. Hinton on Firing Line: AI Is Making ‘Alien Beings’ and Self-Regulation Is ‘Absurd’