UK regulator flags risks from powerful AI models disruptioning finance sector

The Prudential Regulation Authority (PRA) of the Bank of England has raised alarms regarding the potential disruptions that advanced AI technologies could inflict on the financial services sector. A recent report by Reuters highlighted specific concerns surrounding systems such as Anthropic's Claude Mythos and ChatGPT 5.5 Instant, which are drawing scrutiny worldwide. Sam Woods, the PRA's chief executive, stated that it is 'reasonable to expect quite significant disruption' as these AI models gain capabilities in pinpointing vulnerabilities within banking infrastructures. He emphasized the urgency for banks to swiftly address security weaknesses, as lingering technical flaws are among the primary contributors to system outages and disruptions. Woods further stressed the importance of maintaining robust cyber hygiene and the necessity for prompt responses to emerging security threats as AI technologies evolve. He indicated that AI-driven defense mechanisms will play an increasingly crucial role in safeguarding organizations against cyber risks. In a related development, Germany's banking regulator BaFin acknowledged the escalating cybersecurity risks associated with advancements in AI. To tackle these issues, BaFin is planning to establish a new division dedicated to assessing the cybersecurity preparedness and risk management of banks and financial institutions. The International Monetary Fund (IMF) has also issued warnings about the growing cyber threats posed by AI. They noted that the global financial system relies on a highly interconnected digital infrastructure, encompassing software, cloud services, and payment networks. The IMF cautioned that cyber risks now increasingly involve correlated failures that could disrupt financial transactions and undermine confidence in the system. In India, Finance Minister Nirmala Sitharaman convened a high-level meeting to address the challenges posed by AI, particularly in relation to systems like Mythos, which remain largely untested. She emphasized the need for innovative solutions to counter emerging threats effectively. In response to these global concerns about AI-driven cybersecurity, OpenAI has launched a new system called 'Daybreak,' designed for cyber defenders to enhance software security. While this initiative may serve as a competitor to Anthropic's Mythos, it also raises questions about potential misuse of such powerful technologies.