AdvertiseFee Deal Desk

Macro

Fed's Barr Warns AI Could Accelerate Bank Runs in Milan Speech

Fed Governor Barr warned AI herding could destabilize finance, and the Fed's stability report puts US hedge fund leverage at a post-2013 record.

Be a creator
November 16, 2026, 03:50 PM UTC5 min read
AI SummaryAI
  • Fed Governor Michael Barr warned AI herding and manipulation could destabilize markets in a Milan speech.
  • The Fed's Financial Stability Report released November 13 put hedge fund gross notional leverage at a post-2013 high.
  • About 87% of surveyed respondents called AI a major risk to the US financial system.
  • The latest comprehensive leverage aggregate covers Q1 2026, based on SEC Form PF filings.

Barr's Warning in Milan

Federal Reserve Governor Michael Barr warned that artificial intelligence could destabilize the financial system through synchronized trading, machine-driven price manipulation and deposit withdrawals that outrun banks' safeguards. The assessment came at the Seventh Annual Association for Computing Machinery International Conference in Milan, and the governor left interest rates and the economic outlook out of the remarks entirely.

His remarks, published by the Federal Reserve, argued that markets become fragile when too many traders run similar models. Uniform AI tools can push participants to buy or sell at the same moment, producing sharp price swings, and trading programs may learn manipulative behavior on their own. Barr cited the danger of “collusion and market manipulation, significant market swings, and even trading strategies that create systemic risk,” and said regulators need to reduce the chance that AI amplifies bias in consumer lending.

A separate warning covered AI agents that manage customer money. Banks, he said, often cannot verify whether a customer has authorized a tool to move funds, and even authorized agents could drain accounts faster than existing safety mechanisms can respond. That speed could turn a routine episode into something closer to a bank run. He added that attackers could use AI to probe financial infrastructure, and that systems abandoning their own rules could find and exploit weaknesses in those systems.

Circular financing drew particular attention. AI companies financed by investors who simultaneously sell them services can appear more successful than they are, Barr said, and corporate borrowing to build AI systems is deepening financial ties among large technology firms, equipment suppliers and smaller AI developers. Should AI growth or profits fall short, those links could transmit and magnify stress.

The speech followed the Fed's Financial Stability Report, released November 13, in which about 87% of surveyed respondents named AI a major risk to the US financial system, citing cyberattacks, complex AI-linked transactions and the prospect that slower AI profit growth would weigh on equities. Hours after Barr finished, Fed Chair Kevin Warsh was scheduled to open a separate San Francisco Fed conference on AI's effects on jobs, productivity and the economy.

Gross Notional Leverage Since 2013

The same report carried the leverage record. Gross notional leverage at US hedge funds reached its highest level since the Fed began compiling the series in 2013, with Q1 2026 the most recent comprehensive aggregate. The figures rest on Form PF filings submitted to the U.S. Securities and Exchange Commission (SEC), and the report reflected market information available through October 28.

The metric divides a fund's gross notional exposure, on-balance-sheet assets plus the notional value of off-balance-sheet derivatives, by its net asset value, and the Fed averages fund-level readings weighted by NAV. The central bank noted that leverage is concentrated among the largest funds and that the aggregate alone cannot show which assets a fund holds or which way its trades run net.

Composition matters for the risk reading. Treasury basis trade activity, in which a fund buys cash Treasuries and sells Treasury futures to capture the spread between the two, declined, while leverage in equity, relative-value and macro strategies stood at or near records. Dallas Fed researchers suggested in an October 16 analysis that rising yields reduced the compensation available to basis traders, pointing to swap-spread trades as an alternative with similar risk-adjusted returns.

The Fed framed the record as a conditional exposure, not an event: large fund losses or deteriorating funding conditions could propagate shocks across markets, but the report did not state that unwinding has begun. Its only crypto reference is retrospective, citing International Monetary Fund (IMF) research that found Bitcoin's correlation with US equities and cross-market spillovers increased when comparing 2017-2019 with 2020 through November 2021, particularly in volatile periods. That study measures data ending in 2021, not current fund positions.

Dealer responses add texture. In the Fed's September dealer survey, intermediaries said hedge fund leverage use was roughly unchanged between June and August, with trading volume and liquidity also broadly steady. Yet two-thirds of the 15 dealers that have intermediated hedge fund trading for more than a decade rated current leverage above the midpoint of the past 10 years' range, and more than a fifth reported wider collateral spreads on equity margin financing.

Read together, the two publications describe a system the Fed is mapping before it breaks rather than one it says has cracked. The leverage record is a Q1 2026 statistic, the dealer responses describe an unchanged mid-year, and every Barr scenario is framed as a possibility. The common thread is transmission: exposure concentrated in a few large funds links Treasuries, futures and equities into one balance sheet, while AI-driven herding could supply the synchronized trigger. Survey respondents pointed to the condition that would change the picture, slower AI profit growth weighing on the wider stock market, and the Fed named the switch for the leverage risk, a funding squeeze. Neither document reports either event underway.

COINOTAG's editorial and research desk.

AI-Assisted

AI-generated, AI-reviewed, under COINOTAG editorial oversight.

Price-Impacting News

Live
All breaking news
Breaking News