• US and European regulators increase scrutiny of AI-related banking risks
• Fed, OCC and FDIC examine autonomous AI agents and human oversight
• ECB and Bank of England emphasize cyber recovery and operational resilience
• Concerns grow over rogue AI, cyberattacks and systemic financial disruption
• Dependence on external technology providers creates concentration risks
• Banks may face higher costs for cybersecurity and recovery infrastructure
• IMF warns that major cyber incidents could undermine financial confidence
• Regulators assess whether existing governance frameworks remain adequate

Banking regulators in the United States and Europe are intensifying scrutiny of artificial intelligence amid growing fears that autonomous systems, sophisticated cyberattacks and technology dependencies could threaten global financial stability.

Supervisory authorities are examining whether banks have adequate safeguards to control increasingly powerful AI applications, with particular attention to cybersecurity, operational resilience and the potential consequences of systems operating without effective human oversight.

According to Reuters, six people familiar with regulatory discussions said authorities were increasingly concerned that advances in frontier AI could expose financial institutions to vulnerabilities requiring substantial investment in cyber recovery capabilities and stronger third-party risk management.

The Federal Reserve, Federal Deposit Insurance Corporation and Office of the Comptroller of the Currency are examining banks' deployment of advanced AI models, including autonomous agents used in cybersecurity and risk management.

Supervisors are seeking information about the extent of human control over these systems, their ability to operate independently and whether institutions rely on technology developed by Chinese providers.

The heightened scrutiny reflects concerns that agentic AI - technology capable of initiating actions and executing complex tasks with limited human intervention - could introduce risks that existing governance frameworks struggle to contain.

Unlike conventional automated systems, autonomous agents may make operational decisions, interact with external platforms and respond dynamically to changing circumstances, potentially complicating accountability and incident containment.

European regulators are similarly increasing their attention to banks' ability to withstand technology failures and recover from major cyber incidents.

The European Central Bank and Bank of England are emphasizing recovery arrangements, including backup infrastructure capable of restoring critical services following severe disruption.

The concerns extend beyond individual institutions to the possibility that interconnected technology dependencies could transmit operational failures across financial markets.

Banks increasingly rely on common cloud infrastructure, external AI developers and specialized technology providers, creating potential concentrations of risk that may prove difficult to manage during widespread incidents.

The Bank of England has previously warned that dependence on a limited number of technology suppliers could amplify cyber vulnerabilities and create single points of failure affecting multiple financial institutions simultaneously.

In its 2026 supervisory priorities, the Prudential Regulation Authority identified AI adoption as both an opportunity for financial innovation and a source of additional operational, cybersecurity and third-party risks.

The authority emphasized that technological advances must not compromise institutional safety and soundness.

The latest regulatory concerns also coincide with warnings from the International Monetary Fund about vulnerabilities associated with AI investment and growing cybersecurity threats.

In its October financial stability assessment, the IMF highlighted the potential consequences of elevated technology valuations, increasing financial leverage and cyber incidents for global markets.

For banks, the emerging supervisory focus could require substantial changes to AI governance, including stronger model validation, continuous monitoring, clearer accountability and more rigorous testing of recovery arrangements.

Financial institutions may also face pressure to demonstrate that autonomous applications can be interrupted or disabled before unexpected behavior causes wider disruption.

Although regulators have not announced comprehensive new AI-specific banking rules, their inquiries indicate that existing supervisory expectations are being applied more intensively to emerging technologies.

The challenge for banks will be balancing the operational efficiencies and commercial opportunities offered by AI against increasingly demanding expectations for resilience, transparency and control.

As autonomous technology becomes embedded in critical financial processes, regulators appear determined to prevent innovation from creating vulnerabilities capable of spreading beyond individual institutions into the wider financial system.