In the wood-panelled halls of Mansion House on 14 July, Andrew Bailey stood before an audience of financiers and policymakers. The Bank of England governor sketched a future in which artificial intelligence becomes the next engine of economic growth, yet warned that its most advanced forms already threaten the security of the systems on which Britain's financial heart relies.
Bailey described frontier AI as likely the next general purpose technology after steam, electricity and computing. Its promise is real, he suggested, touching on everything from productivity gains to fresh questions about where data centres draw their power and who bears legal responsibility when autonomous agentic systems make decisions. But the tone shifted when he reached the risks. Frontier models, he argued, can uncover and exploit vulnerabilities at a speed and scale that outpaces traditional defences, raising the spectre of cyber attacks capable of shaking financial stability itself.
Cross-border threats demand measured cooperation
Bailey's central prescription was clear. No single country, he said, can seal itself off from systems that cross borders by design. He called for much stronger international coordination on testing frontier models before they circulate widely, beginning with public authorities and extending across critical national infrastructure. The United Kingdom, he noted, enters these conversations from a position of strength thanks to its world-leading AI Security Institute and the established expertise of the National Cyber Security Centre.
We’ve got to get better international understandings of how we deal with the introduction of frontier AI.
That line, delivered in the speech titled Growth and regulation, carried an implicit recognition of limits. Global forums can share threat intelligence and agree on baseline testing protocols, yet they risk producing one-size-fits-all rules that blunt the very competitive advantages Britain has cultivated in AI research, deployment and financial services. Conservative voices in industry have long cautioned against ceding too much ground to supranational bodies whose priorities may favour uniformity over the practical innovation and individual liberty that have powered UK tech clusters from Cambridge to London.
Bailey himself balanced the ledger. He spoke of opportunities alongside dangers, of the need for robust recovery mechanisms such as bare metal restoration that allow systems to be rebuilt from scratch after an incident. Public policy, regulation and international cooperation would all be required, he added, but each must be calibrated to protect rather than constrain British economic interests.
Parliamentary scrutiny turns to bubbles and jobs
Later the same day, Bailey appeared before Parliament’s Treasury Committee. There, alongside other officials, he fielded questions on financial stability and the creeping influence of artificial intelligence across markets, regulation and the wider economy. One exchange stood out. The governor warned that the bursting of an AI stocks bubble would send ripples through the United Kingdom, potentially forcing a monetary policy response in the form of interest rate adjustments.
