IMF warns new AI models risk ‘systemic’ shock to finance - FT中文网
登录×
电子邮件/用户名
密码
记住我
请输入邮箱和密码进行绑定操作:
请输入手机号码,通过短信验证(目前仅支持中国大陆地区的手机号):
请您阅读我们的用户注册协议隐私权保护政策,点击下方按钮即视为您接受。
商业快报

IMF warns new AI models risk ‘systemic’ shock to finance

Fund says preparations needed for ‘inevitable’ AI-enabled breaches of financial institutions’ cyber defences
00:00

{"text":[[{"start":7.4,"text":"The latest AI models pose a risk of “correlated failures” that could affect the financial system on a “systemic” level, the IMF has said, urging policymakers to prepare to deal with an “inevitable” breach."}],[{"start":20.65,"text":"The fund warned that the capabilities of new AI models “elevate cyber risk to a potential macro‑financial shock”."}],[{"start":27.65,"text":"The IMF’s alert underlines how regulators are becoming increasingly alarmed at the potential for Anthropic’s Claude Mythos and other AI models developed by US tech companies to threaten the world banking system by exposing weaknesses in lenders’ cyber defences. "}],[{"start":44.9,"text":"“Advanced AI models can dramatically reduce the time and cost needed to identify and exploit vulnerabilities, raising the likelihood of simultaneously discovering and targeting weaknesses in widely used systems,” senior IMF officials wrote in a blog post published on Thursday."}],[{"start":64.1,"text":"“As a result, cyber risk is increasingly about correlated failures that could disrupt financial intermediation, payments and confidence at the systemic level,” they said, adding that the recent controlled release of Mythos “underscored how quickly risks are increasing”."}],[{"start":81.1,"text":"Anthropic said last month Mythos had “found thousands of high-severity vulnerabilities, including some in every major operating system and web browser”. It added: “The fallout — for economies, public safety and national security — could be severe.” "}],[{"start":96.25,"text":"It plans to roll Mythos out gradually, following its release to a small group of 40 organisations that are mostly US-based, allowing them to fix the vulnerabilities it finds. This includes Amazon and Microsoft as well as large banks such as JPMorgan Chase. "}],[{"start":112.35,"text":"The limited release has enabled some companies to receive more “patches”, technical fixes that close vulnerabilities found by Mythos. But many non-US banks and financial groups have been left without access to the new AI model, raising concerns about uneven levels of protection."}],[{"start":128.6,"text":"Companies that have access to the San Francisco group’s new tool told the FT recently that joint action “across the public and private sectors” was essential to support hospitals, banks and utilities vulnerable to the threats it uncovered."}],[{"start":143.54999999999998,"text":"In the IMF’s first publication about the increased cyber security threat from the latest AI models, it urged policymakers to improve international co-operation. "}],[{"start":154.39999999999998,"text":"“Cyber risk does not respect borders,” it said, adding: “Emerging and developing countries, which often have more severe resource constraints, may be disproportionately exposed to attackers targeting regions with weaker defences.”"}],[{"start":168.24999999999997,"text":"The IMF said: “Attacks become more dangerous when discovery and exploitation scale rapidly, with implications for financial stability.”"}],[{"start":177.19999999999996,"text":"Raising doubts over whether the financial system could withstand a major cyber attack powered by the latest AI technology, the IMF said: “Confidence effects, payment disruptions, liquidity strains and fire‑sale dynamics could follow if multiple institutions are affected simultaneously.”"}],[{"start":195.39999999999995,"text":"With many financial institutions using the same software and shared service providers, the blog said AI models could “create simultaneous vulnerabilities across many institutions”. "}],[{"start":206.99999999999994,"text":"Financial software is “harder to target than open-source infrastructure”, the IMF said. But it added that this mitigation was “likely to erode quickly as model training expands, capabilities diffuse and leaks occur”."}],[{"start":220.84999999999994,"text":"“Defences will inevitably be breached, so resilience must also be a priority, specifically to limit how far incidents spread and ensure rapid recovery,” it said. "}],[{"start":231.64999999999995,"text":"Cyber stress testing, scenario analysis and board-level oversight of cyber risks are “indispensable” to defend the financial system, the IMF said, along with better public-private collaboration on threat intelligence and incident response."}],[{"start":253.84999999999994,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1778204064_6786.mp3"}

版权声明:本文版权归FT中文网所有,未经允许任何单位或个人不得转载,复制或以任何其他方式使用本文全部或部分,侵权必究。

从壮志凌云到“幽灵蝙蝠”:未来空战,打法变了

未来,战斗机飞行员或许不再亲自上阵厮杀,而是坐镇空中、充当任务指挥官,由自主“协同作战飞机”配合作战。

Lex专栏:SpaceX的情况进一步证明季度财报并非必要

马斯克旗下这家火箭制造商过去三个月的财务状况如何,其实无关紧要。

谷歌为Anthropic打造的2000亿美元华尔街融资机器

私募信贷、芯片租赁和数据中心担保,支撑起AI支出的全新庞大模式。

“日元干预”等于“美国自保”

美国联手日本支撑日元,不只是出于盟友情谊,更是为了避免日本加息或美国国债遭抛售、导致美债收益率进一步走高。

“诅咒之岛”:科技游民与诈骗犯藏身的千亿美元奢华开发项目

警方的突击搜查再次打击了马来西亚陷入困境的中资“森林城市”项目的声誉。

问题不在因凡蒂诺

马杜罗:应该将国际足联的监管职能与商业活动分开,其治理应真正做到包容并具有代表性,监督必须真正独立。
设置字号×
最小
较小
默认
较大
最大
分享×