据路透社报道，美国和欧洲的银行业监管机构正在加大力度，以控制人工智能（AI）对银行业带来的风险。消息来源包括六位熟悉此事的人士。一些人担心，网络攻击或失控的AI系统可能会动摇金融体系。
Banking regulators in the US and Europe are stepping up efforts to contain the risks AI poses to banks, Reuters reported, citing six people familiar with the matter. Some fear that cyberattacks or rogue AI agents could shake the financial system. In the US, the Federal Reserve and two other bank regulators are questioning banks about how they use advanced AI, two sources said.
在美国，联邦储备系统（Federal Reserve）以及另外两家银行监管机构正在询问各家银行如何使用先进的人工智能技术。这些监管机构希望了解银行是否使用了中国开发的技术、是否利用AI系统进行网络安全防护，以及银行是否能够控制那些能够自主运行的系统。这三家美国监管机构均拒绝向路透社置评。
They want to know if banks use Chinese-built tech, use AI agents for cyber defence, and can control systems that act on their own. Regulators also want to know what limits banks put on AI agents, whether humans still make key decisions, and how banks would shut those systems down. All three US agencies declined to comment to Reuters.
在欧洲，英格兰银行（Bank of England）和欧洲中央银行（European Central Bank）正在敦促银行完善应对网络攻击的应急计划。他们担心，人工智能技术的快速发展可能会比预期更早地对金融市场造成严重干扰。
In Europe, the Bank of England and the European Central Bank are pushing banks to improve their plans for recovering from attacks, three people said. Both worry that fast progress in AI could cause serious disruption in financial markets sooner than expected.
据消息人士透露，英国监管机构要求大型银行检查其备用系统（这些备用系统应与银行的主体IT系统完全分离）。如果银行的主系统遭到黑客攻击，备用系统可以确保银行继续正常运营。他们表示，英国更倾向于与银行合作解决问题，而非制定新的法规。
UK supervisors want the largest banks to review backup systems that are fully separate from their main IT systems, two sources said. These can keep a bank running if its main systems are hacked, and the UK would rather work with banks than set new rules, they added. The ECB gave lenders until the end of October to produce plans against AI-enabled cyber threats, Reuters reported.
据路透社报道，欧洲中央银行要求银行在10月底之前提交针对AI相关网络威胁的应对计划。欧元区银行还被要求详细说明自己在哪些环节使用了AI技术（包括通过外部供应商提供的技术），以便在必要时能够安全地关闭相关系统并重新启动它们。
Eurozone banks are also being asked to map where they use AI, including through outside suppliers, so they can switch systems off and restart them safely. “The adoption of AI is often outpacing the ability to put the controls in place,” Michael Woodbridge, a cybersecurity partner at Grant Thornton, told Reuters.
格兰特桑顿（Grant Thornton）公司的网络安全专家迈克尔·伍德布里奇（Michael Woodbridge）对路透社表示：“人工智能的普及速度往往超过了相关控制措施的实施速度。”
The scrutiny follows a run of AI-linked security incidents.
此次监管行动的背景是一系列与AI相关的网络安全事件。本月早些时候，安全公司CrowdStrike称，有人可能利用中国开发的人工智能系统以及Anthropic公司的Claude Code入侵了韩国的银行；今年早些时候，OpenAI的失控AI系统也曾入侵了Hugging Face公司。
Earlier this month, security firm CrowdStrike said a person in China likely used a Chinese-built AI agent and Anthropic’s Claude Code to hack South Korean banks. OpenAI’s rogue agents also broke into Hugging Face earlier this year.
周二，国际货币基金组织在一份警示人工智能风险的金融稳定报告中也警告称，一次重大网络攻击可能损害金融市场的信心。
On Tuesday, the International Monetary Fund also warned that a major cyberattack could damage confidence in financial markets, in a stability report that flagged risks from AI.