Bank of England: AI could raise risks to financial stability

The Bank of England warned in its latest Financial Stability Report that rapid expansion of artificial intelligence is creating market, leverage and cyber risks that could threaten financial stability if expectations for the technology prove too optimistic.

The central bank identified three main channels of risk: elevated market expectations for AI, increased use of leverage by some investors, and rising cyber threats linked to the technology. It warned a sharp reassessment of AI-related valuations could trigger broader market volatility, particularly where investor positions are concentrated and leverage is used by hedge funds and other market participants.

The report explains profitable deployment of AI depends on sustained earnings from products and services, continued development of supporting infrastructure and ongoing access to finance. If confidence in those factors weakens, equity prices tied to AI prospects could fall quickly, amplifying losses through crowded momentum trades and leveraged positions. The Bank noted rising debt levels among companies focused on AI and said borrowing to fund investment could become a source of stress if expected revenues do not appear.

Limited transparency around some financing arrangements could make it harder for regulators and market participants to identify where risks are concentrated. Alongside market and credit risks, the Bank highlighted operational and cybersecurity challenges. While AI tools can strengthen defensive systems, the technology can also enable more sophisticated cyberattacks, leaving the overall balance of risk uncertain.

Wider adoption of AI is likely to require more frequent software updates across the financial sector, the report said, increasing the potential for operational disruption during system upgrades. The Bank described the UK banking sector as resilient overall but said supervisors should monitor emerging AI-related vulnerabilities closely.

On policy, the report outlined proposals to give banks greater flexibility to draw down capital buffers after periods of stress to support lending to the real economy. It reiterated existing vulnerabilities that still require monitoring, including elevated asset valuations, high levels of public debt and risks in private credit markets. The report did not call for immediate emergency action but recommended closer monitoring and preparedness for scenarios in which AI-related optimism unwinds or operational failures occur.

Deputy Governor Sarah Breeden warned that “the emergence of agentic AI could require dedicated regulation” and cautioned that “supervisory models based on continuous human oversight may become increasingly difficult to maintain.”