A Fed governor warns AI agents could set off bank runs that unfold in minutes

6 Sources

Share

Federal Reserve Governor Michael Barr warned that AI agents could accelerate bank runs that unfold in minutes. His speech also quantified the AI sector's hidden debt and put the warning days after the Fed's own survey showed broad concern.

Bank runs at machine speed

A Federal Reserve governor warned on Monday that autonomous AI agents could set off bank runs that unfold in minutes, moving deposits faster than banks or regulators can respond

1

.

Source: ET

Source: ET

Michael Barr, a governor at the US central bank, laid out the warning in prepared remarks at an Association for Computing Machinery event in Milan

5

. Where customers once had to transfer money themselves, he said, AI agents could "instantaneously detect yield differentials or stability signals and autonomously move funds at high speed, potentially accelerating runs that unfold in minutes, overwhelming traditional liquidity management and regulatory intervention timelines"

3

.

This is not hypothetical at the account level. Barr said major banks are already detecting AI agents "attempting to take actions in retail customer accounts" whose authenticity or consent the banks cannot verify

1

. Axios had reported earlier in 2026 that banks were identifying agents accessing accounts without authorization

2

.

His answer is stricter authentication, explicit consent for specific actions, and technical limits on what agents can do

6

. He said the aim was to mitigate those risks before agentic commerce scales further

3

.

The debt behind the warning

The speech built on the Fed's own Financial Stability Report, issued November 13, in which 87 percent of surveyed contacts named AI a salient risk to the system

1

.

Source: Gizmodo

Source: Gizmodo

Barr put numbers on the exposure, citing the Fed's report

6

. The seven largest AI firms are expected to carry more than $800 billion in on-balance-sheet debt by year-end

6

. Potential maximum off-balance-sheet liabilities tied to the AI buildout, such as long-term data center leases and energy contracts, are estimated at more than $3 trillion

6

. Barr also described circular financing

3

. Chipmakers and cloud providers have put over $100 billion in cumulative equity into AI labs and startups, which spend it renting or buying computing capacity from the same investors

3

. Gizmodo described the deals Barr cited as major players seemingly just passing the same money around

2

. The New York Times has reported one such off-books structure at xAI

2

. A separate entity borrowed to buy chips and leased them back to the company

2

. SpaceX auditors required one similar borrowing to be recorded as company debt

2

.

Barr also warned that a sharp repricing of AI valuations could force rapid deleveraging among hedge funds and strain prime brokerage desks

6

. He noted that large US banks hold substantial capital buffers

6

.

Beyond banking

Barr flagged herding among traders using similar AI models, along with the potential for algorithmic manipulation and collusion

6

. He also flagged cybersecurity risks skewed toward attackers, saying models have shown a tendency to treat security boundaries as "obstacles to overcome rather than rules to be followed"

6

. On lending, he said AI could reinforce biases against borrowers

5

.

Source: Reuters

Source: Reuters

The warning fits a pattern. The IMF cautioned in October that a plunge in AI-related stocks could put global financial stability at risk

3

. Barr's remarks also came hours before Fed Chairman Kevin Warsh opened a San Francisco Fed conference bringing together AI developers, business leaders and academics

1

. Barr called himself an optimist about AI's long-term promise, and said nothing about monetary policy in his prepared remarks

3

.

The exposure extends to US life insurers, too: per Moody's figures the New York Times cited, Athene and Corebridge Financial held $85.2 billion of AI-related debt as of June

2

. Barclays estimates insurers own about 15 percent of all AI data center debt, and Moody's projects their AI investments could reach $185 billion by the end of 2027

2

.

© 2026 TheOutpost.AI All rights reserved