Global financial regulators are calling for tighter oversight of increasingly autonomous artificial intelligence systems as their use expands across the banking and financial services sector.

The Financial Stability Board (FSB), an international body responsible for monitoring and making recommendations about the global financial system, has warned that advanced forms of AI could introduce new risks if not properly governed.

In a report released on Wednesday, the FSB strongly encouraged financial institutions to implement safeguards around the use of “agentic AI” — a new generation of AI systems capable of planning, reasoning, and carrying out tasks with minimal human intervention.

Agentic AI gaining momentum in finance

Agentic AI is already being deployed by banks and financial firms in areas such as:

  • Fraud detection
  • Customer service operations
  • Compliance monitoring
  • Back-office administration
  • Workflow automation

According to research from the Cambridge Centre for Alternative Finance, more than half of surveyed financial institutions reported active use of agentic AI technologies.

The study found that:

  • 52 per cent of respondents are already using agentic AI
  • 23 per cent have deployed the technology at scale or are undergoing major transformation projects
  • 29 per cent are currently testing or piloting agentic AI applications

The figures highlight the rapid pace at which financial organisations are integrating autonomous AI systems into critical business functions.

Regulators warn of new risks

The FSB cautioned that agentic AI introduces unique challenges compared with traditional software systems.

Because these AI systems can make decisions and take actions independently, regulators fear they may:

  • Execute unauthorised actions
  • Trigger unexpected financial transactions
  • Create cybersecurity vulnerabilities
  • Cause data breaches
  • Disrupt interconnected financial systems

The report noted that risks associated with autonomous AI can materialise quickly, potentially leaving organisations with little time to intervene once problems emerge.

One of the key concerns highlighted by regulators is the challenge of maintaining effective human oversight.

The FSB warned that AI agents may pursue actions that differ from an organisation’s intended objectives without employees immediately noticing or being able to stop them.

New safeguards proposed

To address these concerns, the global watchdog has proposed a set of non-binding best practices aimed at helping financial institutions manage AI-related risks.

Among the recommendations are:

  • Establishing clear limits on what AI systems are permitted to do
  • Implementing safeguards around autonomous decision-making
  • Requiring human approval for high-risk activities
  • Introducing monitoring mechanisms to track AI behaviour
  • Creating governance frameworks specifically designed for agentic AI

The report specifically recommends that financial transactions exceeding predetermined thresholds should continue to require human authorisation.

AI agents could be treated like employees

One of the more notable suggestions in the report is that organisations may need to adapt human resources policies and governance structures to account for AI systems.

The FSB said firms could consider treating autonomous AI agents as “synthetic employees,” applying similar oversight, accountability measures, and operational controls used for human workers.

This approach reflects the growing role AI is playing in business operations and acknowledges that autonomous systems may increasingly perform functions traditionally handled by employees.

Industry feedback period now open

The FSB has opened its proposed guidelines for public consultation, with feedback being accepted until July 22.

As financial institutions continue investing heavily in artificial intelligence, regulators worldwide are seeking to strike a balance between encouraging innovation and protecting the stability, security, and integrity of the global financial system.

The latest warning underscores a growing consensus among regulators that while AI offers significant opportunities for efficiency and growth, stronger governance will be essential as increasingly autonomous systems become embedded within critical financial infrastructure.