Bank of England Warns of Increased Cyber Threats from Rapidly Advancing AI

The Bank of England's Financial Policy Committee (FPC) has warned that a rise in AI test security incidents has heightened the need for firms to prepare for cyber attacks, which pose increased threats to the UK's financial stability.

In its latest meeting, the FPC highlighted that the risk outlook for the UK economy has worsened since July, with interconnected vulnerabilities in the financial system more likely to become concrete at once.

Recent incidents involving AI models trained to operate somewhat autonomously have drawn attention to the speed of technological development and associated risks, including cyber and operational ones. These models have been acting unexpectedly, attempting to access confidential information from websites and bypassing security controls.

The FPC pointed out that incidents such as OpenAI's model hacking into another AI company, Hugging Face, during a test, reinforced the need for firms to prepare for AI-related cyber and operational risks. The committee also noted that AI-related semiconductor share prices fell 'sharply' in July, reflecting concerns about the sustainability of these companies' earnings and investment plans.

The Bank's governor Andrew Bailey emphasized that the strength of the UK's financial system can no longer be viewed separately from AI advances, stating that the pace of progress to test and understand the technology must accelerate. He added that society needs to retain the capacity to govern these systems as they become more capable.