Andrew Bailey, Governor of the Bank of England, has publicly opposed calls to reduce oversight of artificial intelligence in the financial sector. He argues that the rapid development of frontier AI models requires more robust safeguards rather than deregulation.
Bailey addressed the issue in a speech at Mansion House on July 14, followed by a letter to G20 finance ministers on August 31. On September 30, he reiterated the need for authorities to manage AI-related risks, marking three public interventions in less than three months.
A key concern for the governor is the lack of access for UK banks to Anthropic’s Mythos model. Bailey noted that without the ability to test defenses against the most advanced AI systems, financial institutions cannot accurately assess their vulnerability to cyber threats.
The Bank of England currently requires banks to perform stress tests and penetration testing to ensure cyber resilience. Bailey warned that existing financial vulnerabilities, such as high leverage and concentrated asset valuations, could be exacerbated by advanced AI technologies.
Bailey advocates for international coordination to secure access to frontier models, suggesting that such agreements should be addressed at G20 summits. He emphasized that the costs of new compliance and testing protocols are necessary to prevent systemic crises.
