The regulatory arm of the Bank of England has sounded a cautionary note about the impact that the latest artificial intelligence models may have on the financial sector.
Sam Woods, chief executive of the Prudential Regulation Authority, told attendees at UK Finance’s Growth Delivery Summit that it is reasonable to expect substantial disruption from advanced AI systems, including Anthropic’s Mythos and ChatGPT 5.5 Instant. He stressed that such models are increasingly capable of identifying system vulnerabilities.
Woods highlighted a specific operational risk: the speed of patching. He said that delays in addressing discovered weaknesses are a principal driver of outages across the financial system, and therefore banks must act more quickly to close gaps when they are exposed.
To reduce risk, the PRA head urged financial firms to raise the baseline of cyber hygiene and to shorten response times to emerging threats. He also noted that defenses powered by AI are becoming an important component of risk management.
The comments reflect concerns raised about the implications of new model releases. Anthropic rolled out its Mythos model to a limited set of businesses in April, a move that cybersecurity specialists have said presents challenges for banks operating with legacy technology stacks.
Separately, a Bank of England co-led cyber group reported last month that the sector was prepared for the challenges posed by these developments. The regulatory message from Woods places emphasis on operational remediation and the adoption of AI-driven defensive tools, rather than introducing new regulatory measures in the remarks delivered at the summit.
Key takeaways:
- Senior PRA leadership expects notable disruption from the newest AI models.
- Patching delays are identified as a leading cause of outages, making rapid remediation a priority.
- Financial firms are urged to upgrade basic cyber hygiene and incorporate AI-enabled defenses.