Bailey warns G20 against threat of AI-related cyber disruption

In a letter to G20 finance ministers and central bank governors ahead of a meeting in the US, Bank of England Governor Andrew Bailey warned that frontier artificial intelligence (AI) models pose significant systemic risks to the global economy.

Writing in his role as chair of the Financial Stability Board (FSB), Bailey cautioned that advanced models exhibit increasingly autonomous problem-solving capabilities alongside significant threat potential. He stressed that these technologies could destabilise the interconnected international financial system through cyber disruption that can rapidly spread across borders.

Bailey raised particular concern that numerous jurisdictions currently lack adequate regulatory protocols to govern the development, release, and deployment of frontier models, thereby intensifying vulnerabilities within the financial sector and across wider industries. He called on authorities responsible for safeguarding global financial architecture to take coordinated international action, urging them to prioritise robust measures that ensure advanced systems are released and deployed safely and responsibly.

The news comes as a survey by Sequence Security found that 65% of organisations have seen AI agents acting out of scope, while another report reveals that OpenAI agents hijacked a German website in a previously undisclosed incident. 

In related news, OpenAI has confirmed that its engineers are developing automated shutdown capabilities for its AI systems and will restrict (£) who can access its recently launched Astra model del due to its advanced cybersecurity capabilities. Meanwhile, Anthropic has confirmed in a blog post that recent cybersecurity incidents are a "failure of operational security" and confirmed that it has tightened testing procedures. 

£ - This article requires a subscription. 

Training Announcement: The BCS Foundation Certificate in AI examines the challenges and risks associated with AI projects, such as those related to privacy, transparency and potential biases in algorithms that could lead to unintended consequences. Explore the role of data, effective risk management strategies, compliance requirements, and ongoing governance of the AI lifecycle and become a certified AI Governance professional. Find out more.