Finnect AI Insights

Policy · · 4 min read

Can AI Really Trigger Bank Runs and Market Chaos, as the US Fed Warns?

Learn how AI's speed and complexity could destabilize financial markets, from rapid trading to accelerated bank runs, and what businesses need to do.

War of wealth bank run poster

In a stark warning from Milan, Italy, US Federal Reserve Governor Michael Barr recently laid out a series of profound risks that Artificial Intelligence (AI) poses to the global financial system.

He wasn't just speculating; his remarks built on a sobering central bank report. That report, the Fed's latest Financial Stability Report, revealed that a striking 87% of financial contacts surveyed see AI as a significant threat. Their concerns range from sophisticated cyberattacks to complex AI-related financing schemes.

Barr expanded on these worries, painting a picture where the very tools designed for efficiency could become instruments of chaos. Think lightning-fast trading algorithms, self-learning models, and even AI agents managing your personal bank account.

But this isn't just another tech-skeptic's lament. Barr's address detailed specific, tangible ways AI could unravel the safeguards traditionally built into our financial infrastructure. It highlights a critical tension: the promise of AI versus its potential to destabilise.

The real question isn't if AI is powerful, but how its power could manifest as systemic risk. That raises three questions.

Question 1

What Specific AI Risks Are Causing Concern at the US Federal Reserve?

Federal Reserve Governor Michael Barr's recent address in Milan outlined a range of perils, from market manipulation to consumer protection issues. The primary concern, underscored by the Fed's Financial Stability Report, is the widespread belief among financial contacts that AI is a salient risk. According to Reuters, 87% of those surveyed cited this.

For context, these concerns aren't theoretical. Barr specifically pointed to the potential for AI models in trading to encourage "herd-like behavior", where many automated systems react similarly and simultaneously, disrupting markets. He warned that self-learning algorithms could actively manipulate and destabilise markets, creating artificial volatility.

Beyond trading, Barr highlighted the difficulty banks face in verifying if an AI agent "attempting to take actions in retail customer accounts" is genuinely authorised by the customer. This lack of certainty opens doors for fraud or unintended consequences. He also stressed the need to prevent AI from reinforcing biases in consumer lending, ensuring fair access to credit. Cyber risks are another major worry, with AI-enabled attackers or even rogue AI agents potentially exploiting vulnerabilities in the financial system, according to CNBC-TV18.

Question 2

How Could AI-driven Trading and Customer Agents Accelerate Financial Instability?

The speed and autonomy of AI are a double-edged sword. Barr warned that even legitimate AI agents, acting at lightning speed for retail banking customers, could accelerate bank runs and overwhelm traditional safeguards. Imagine a scenario where a rumour or a minor market tremor triggers thousands of AI agents to simultaneously initiate autonomous transfers.

For context, a traditional bank run involves many customers physically withdrawing money or transferring funds over hours or days. This gives banks and regulators time to react. But with AI, as Barr noted, "autonomous transfers could drive bank runs unfolding in minutes." This could overwhelm banks' liquidity management and regulators' response times.

Think of it like a fleet of automated delivery bikes (illustrative number: 10,000 bikes) that suddenly receive a simultaneous order to pick up and deliver goods from one specific warehouse, all at the exact same minute. The warehouse's loading docks, staff, and road access points (the bank's liquidity and operational capacity) would be instantly overwhelmed, causing a system-wide jam. It's not about malicious intent, but the sheer, synchronised speed that breaks the system.

This rapid, collective action by AI could create "systemic risk" (a risk that affects the entire financial system, not just one firm), leading to significant market volatility, as Barr stated. And that brings us to the third question.

Question 3

What Does the US Fed's Warning Mean for Businesses Using or Developing AI?

The Fed's concerns translate into clear imperatives for businesses. Barr stressed that AI developers often struggle to ensure their models are "well controlled, numerically and legally precise, explainable, and replicable." For businesses deploying AI, this means a rigorous focus on governance, transparency, and accountability.

For context, this isn't just about avoiding a crisis; it's about building trust and ensuring regulatory compliance. Barr called for "strict authentication, consent checks for specific actions, and technical limits on AI agents." This implies that every AI action, especially those involving customer funds, needs clear audit trails and explicit user approval mechanisms, even if automated.

He previously warned in February 2025 that competitive pressures to adopt generative AI could "heighten risks in financial services." The very attributes that make AI attractive, speed, automation, and optimisation, also present risks like increasing market volatility and stoking asset bubbles. This means businesses need to balance the pursuit of efficiency with robust risk management frameworks. It's about designing AI systems that can explain their decisions, operate within defined boundaries, and be easily audited, rather than being black boxes.

The Bigger Picture

The advent of AI offers transformative potential for every sector, from manufacturing to healthcare. Yet, as the US Fed's warnings highlight, this power comes with profound responsibilities, especially in critical sectors like finance. The story of AI in business will be defined not just by its capabilities, but by our ability to control and understand its consequences.

Businesses must proactively build safeguards, ensuring that the pursuit of efficiency doesn't inadvertently lead to systemic fragility. The future of AI will depend on how well we design these intelligent systems to be both powerful and predictably safe.

It's about making sure that while AI can run, it doesn't run away.

Quick Answers

What are the main AI risks identified by the US Federal Reserve?

Federal Reserve Governor Michael Barr highlighted risks including herd-like trading behaviour, market manipulation by self-learning algorithms, difficulty verifying AI agents, potential for accelerated bank runs, and increased cyberattack vulnerabilities from AI-enabled attackers.

How could AI accelerate a bank run?

AI agents acting at lightning speed for retail customers could initiate autonomous transfers simultaneously. This rapid, collective action could overwhelm a bank's liquidity management and regulators' response times, causing a bank run to unfold in minutes rather than hours or days.

What does the Fed's warning mean for businesses using AI?

Businesses must prioritise strict authentication, consent checks for specific actions, and technical limits on AI agents. They also need to ensure AI models are well controlled, explainable, and replicable to manage risks like market volatility and biases in consumer lending.

Sources

Facts are credited to the outlets above. The analysis, calculations and Signal Check are by the Finnect AI Insights desk.