Bank of England says it is testing AI risks to financial system
Summary (EN)
Reuters reported on April 16 that the Bank of England is actively testing AI-related threats to the financial system through scenario analysis and simulations, according to a letter published by lawmakers. The central bank said it disagreed with criticism from the Treasury Committee that it had adopted a passive wait-and-see stance, and instead argued that it is examining how AI investment and adoption could change market structure and financial stability. Deputy Governor Sarah Breeden said the work includes coordination with international counterparts to understand how AI agents might affect trading behavior, with particular attention on herding effects that could intensify market selloffs during stress events. Reuters linked the timing of the statement to rising concern after Anthropic’s Mythos launch, which has sharpened debate about cyber and operational risks from highly capable AI systems. Bank of England Governor Andrew Bailey had already warned earlier in the week that the new model might open up a broader cyber-risk problem. The Reuters report also highlighted political pressure on the UK Treasury over whether large AI and cloud providers should be brought more quickly into the Critical Third Parties regime that oversees essential financial-system infrastructure suppliers. This development is notable because it moves the AI discussion beyond firm-level experimentation and into macroprudential supervision. Within the past 24 hours, it stands out as an important regulatory signal that central banks are beginning to treat advanced AI and agentic behavior as potential systemic-risk variables rather than merely operational technology issues.
Summary (ZH)
路透社于 4 月 16 日报道，英国央行正在通过情景分析与模拟测试评估 AI 对金融体系带来的风险。这一信息来自提交给议会委员会的一封公开信。英国央行在信中回应了财政委员会对其“观望式”态度的批评，强调自己并非被动等待，而是在持续研究 AI 投资和应用如何改变金融体系结构与稳定性。负责金融稳定事务的副行长 Sarah Breeden 表示，央行正与国际同行合作，评估 AI 代理可能如何影响金融市场交易行为，特别关注在市场压力时期可能出现的“羊群效应”，即算法和代理行为趋同导致抛售被进一步放大。路透社指出，这一表态发生在 Anthropic Mythos 发布引发更大范围讨论之后。由于该模型具备强大的编码与漏洞发现能力，网络安全和系统性运行风险的担忧显著升温。英国央行行长 Andrew Bailey 本周早些时候也已公开表示，新模型可能把整个网络风险问题“打开”。报道还提到，英国财政部正面临议会压力，要求更快决定是否将大型 AI 和云计算公司纳入关键第三方监管框架。过去 24 小时内，这一动态的重要性在于，它标志着 AI 风险讨论已经从单个机构的技术试验层面，进入到中央银行和宏观审慎监管的系统稳定框架之中。