Bank of England Governor Andrew Bailey has stated that formal regulation of artificial intelligence is “not the right place to start,” arguing instead for a framework of “rigorous testing” to manage the technology’s rapid integration into the financial system.
In a shift toward what he described as interventionist testing, Bailey emphasized on September 30, 2026, that advanced AI models must undergo scrutiny both before and after deployment. This approach is intended to identify vulnerabilities and establish “credible points of intervention” rather than attempting to regulate the underlying architecture before its failures are fully understood.
The Governor’s comments come as the financial scale of the AI sector creates new pressures on global stability. Between January and September 2026, AI companies issued $450 billion in debt. This figure significantly exceeds the UK government’s planned gilt issuance for the entire 2026 fiscal year, which stands at approximately $333 billion (£252 billion). This concentration of corporate debt has drawn the attention of central bankers concerned about market exposure to a single, rapidly evolving sector.
The Risk of ‘Recursive Learning’
A primary technical concern cited by Bailey is the emergence of “recursive learning.” This occurs when AI systems progressively govern themselves in a “closed loop,” a process that can systematically reduce human oversight. According to Bailey’s analysis of frontier AI governance, these autonomous feedback loops present a unique challenge to institutional control, potentially making traditional regulatory oversight ineffective if the systems move beyond human-defined parameters.
The call for testing over direct regulation is also informed by recent security failures in the industry. Reporting indicates that OpenAI recently paused testing of a frontier model following a breach of Australia’s public healthcare system. Such incidents highlight the volatility of model deployment and support the Bank’s preference for identifying specific “intervention points” rather than applying broad legislative constraints.
Financial Policy Committee Outlook
The Bank of England’s Financial Policy Committee (FPC) recently updated its risk outlook for the UK economy, noting a deterioration since July 2026. In its September 25 meeting record, released alongside Bailey’s statements, the FPC identified the combination of AI-related debt and escalating geopolitical conflict as key contributors to a more fragile financial environment.
While the AI debt bubble remains a central pillar of this risk assessment, the FPC also noted that energy-linked financial stability is under pressure due to re-escalated conflict in the Middle East. The committee suggested that the speed of AI adoption, coupled with high levels of leverage, necessitates a more agile response than traditional rule-making allows.
By prioritizing testing, the Bank of England aligns itself with the ongoing work of the UK’s AI Security Institute. The strategy aims to allow the technology to develop while ensuring that the central bank maintains the capability to step in should the $450 billion debt pile or autonomous model behavior threaten wider economic stability.
