Financial Regulators Admit AI Trading Systems Outpace Human Oversight

Regulators Acknowledge a Blind Spot

Global financial regulators have publicly acknowledged what many in the industry already knew: the artificial intelligence systems now making decisions about loans, insurance claims, and investments operate beyond the practical reach of real-time human oversight.

The Financial Stability Board, which sets policy direction for banking regulators worldwide and is chaired by Bank of England Governor Andrew Bailey, released a consultation report on June 10 that contains a striking admission. When discussing "agentic AI"—systems that autonomously pursue goals across hundreds or thousands of steps—the FSB states plainly that "monitoring and detecting these actions in real-time can be very difficult."

The report goes further, warning that when these systems err, "overriding, redressing, or remediating these actions can be difficult or impossible for humans." This represents a fundamental shift from the traditional model where qualified personnel directly oversee automated decision-making.

According to Forbes, which first reported the details, the FSB describes an "impracticality of real-time human monitoring of agent decisions as their use scales," potentially leaving staff unaware of system actions or unable to intervene promptly.

The Proposed Solution: More AI

The FSB's response to this oversight gap is itself noteworthy. The report suggests that effective monitoring of autonomous AI agents "may require augmentation with another AI agent or other forms of AI." In essence, regulators are proposing that machines monitor machines.

This approach is already gaining traction. The International Organization of Securities Commissions finalized its own AI supervisory toolkit on May 25, noting that some authorities are experimenting with AI systems that assess and oversee other AI—a practice the industry has dubbed "AI as a judge."

The FSB emphasizes that human accountability remains paramount, with boards and senior managers retaining "ultimate accountability" for AI decisions. However, the practical reality is shifting: humans are moving from direct oversight of decisions to setting boundaries and parameters for AI systems.

Already Deployed at Scale

These systems are not theoretical. The FSB report includes a case study of a large international bank that built an AI agent to detect fraud across more than 80 million signals daily, covering card payments, transfers, and online banking. Developed in-house in three months, the system now proposes most of the bank's card-fraud rules and helped reduce fraud losses by over 20 percent in early 2026 compared to the same period in 2025.

The success story illustrates the paradox: the agent operates at a scale no human team could match, yet the same report acknowledges no human team could fully monitor it either. While this particular bank maintains human approval for new rules, the FSB warns this model becomes impractical as agent deployment scales.

Beyond fraud detection, these autonomous systems are spreading into credit assessments, customer service, trading workflows, and compliance reviews—the infrastructure that determines who receives loans, how claims are processed, and what investments are made.

Why it matters

This shift creates accountability gaps with real consequences for consumers and systemic stability. When AI systems make opaque decisions about financial products, customers may struggle to understand or challenge outcomes. More broadly, if most institutions rely on similar AI models and cloud infrastructure, the FSB warns of "correlated behaviours across financial institutions, amplifying risks like herding and procyclicality"—meaning a crisis could spread faster as look-alike systems react identically.

The timing concern is equally significant: the FSB document is a consultation draft, open for comment until July 22, with final guidance not expected until October. These are recommendations, not binding standards. Meanwhile, the AI agents are already operational and making decisions that affect millions of accounts.

What This Means for Institutions and Individuals

For financial institutions, the regulatory direction is clear even if not yet mandatory: prepare for AI-monitoring-AI architectures and ensure governance frameworks can handle systems that operate faster than human review cycles allow.

For individuals, Forbes notes several practical steps: request written explanations for AI-driven decisions and ask whether humans reviewed them, question how AI-powered services check and correct their decisions, and monitor which firms rely heavily on the same AI suppliers, as concentration risk concerns regulators most.

The fundamental bargain of financial services—that competent oversight protects customer interests—remains intact in principle. In practice, that oversight is increasingly performed by algorithms watching other algorithms, with humans positioned further from direct decision-making than traditional models assumed.

The details of this regulatory acknowledgment were first reported by Forbes contributor Dara-Abasi Ita.