字号·· | 护眼
彭博社

美国科技故障因何而起,专家为何预计会有更多What Caused the US Tech Outages and Why Experts Expect More

点「原文对照」整页切到原文,或双击某段只看那段的原文。

11月13日,在弗吉尼亚州阿灵顿的罗纳德·里根华盛顿国家机场,一名旅客在全国性停飞期间坐在行李传送带上。A traveler sits on a baggage carousel during a nationwide ground stop at Ronald Reagan Washington National Airport in Arlington, Virginia on Nov. 13.

发布时间:2026-11-20T23:51:37+08:00 一系列始于11月11日的技术故障在美国引发了集体恐慌。导致微软公司软件瘫痪、航空出行陷入混乱、国家安全局工作受阻的故障,促使消费者抢购应急物资,仿佛这些是即将到来的大灾难的征兆。网络安全专家表示,这确实是未来趋势的一个信号,因为老旧技术越来越需要被替换。每起事件的具体情况各不相同。

Published: 2026-11-20T23:51:37+08:00 A series of technology outages that began on Nov. 11 spawned a collective freak-out across the US. Breakdowns that knocked out Microsoft Corp. software, snarled air travel and disrupted the work of the National Security Agency prompted consumers to panic-buy emergency supplies as if these were signs of an impending cataclysm. Certainly, cybersecurity experts say, they were a sign of things to come, as old technology increasingly needs to be replaced. The specifics in each case varied.

微软表示,由于一个漏洞导致在原本例行的用户流量重新路由过程中发生系统故障,该公司遭遇了异常漫长的瘫痪期。据美国官员称,美国空中交通经历了几十年来最糟糕的一个周末,因为一名承包商试图在阿肯色州一个偏远地点更换过时线路,导致一台路由器压垮了空管网络上的其他技术设备。

Microsoft said it was hobbled for an unusually extended period because a bug caused a system malfunction during an otherwise routine rerouting of user traffic. US air traffic experienced the worst weekend in decades because a contractor tried to replace outdated wiring at a remote location in Arkansas, according to US officials. As a result, a router overwhelmed other tech on the air traffic control network.

摩根大通公司表示,11月16日遭遇的交易延迟源于本地部署硬件系统的一个软件漏洞,并补充称该问题与网络安全无关。美国国家安全局人员之间的电子通讯在一套机密计算机网络崩溃后中断,联合航空因与老化基础设施相关的设备问题延误了一些航班。苹果公司不得不在一次内部更新出错后恢复部分员工工具、零售业务和消费者服务。

JP Morgan Chase & Co. said that a software bug in on-premise hardware systems was at the root of transaction delays it experienced on Nov. 16, adding that the issue wasn’t related to cybersecurity. Electronic communications among NSA personnel stopped when a classified computer network tanked, and United Airlines delayed some flights because of equipment problems related to aging infrastructure. Apple had to restore some employee tools, retail operations and consumer services after an internal update went awry.

这些后果较轻的故障之所以无疑获得了额外关注,是因为它们发生的时间点。支持关键互联网基础设施的非政府组织“分组清算所”秘书长比尔·伍德科克表示,此类中断可能看起来呈集群式出现,“因为大多数时候人们并不关注它们,但随后某一起事件引起了公众注意,人们便开始到处看到它们了”。

The less consequential of these outages undoubtedly got extra attention because of their timing. Such disruptions can appear to come in clusters “because most of the time people aren’t paying attention to them, but then one gets the public’s attention and people start seeing them everywhere,” said Bill Woodcock, the secretary general of Packet Clearing House, a nongovernmental organization that supports critical internet infrastructure.

“没有任何证据表明它们有关联,”风险情报公司 Dataminr 的威胁研究与网络工程总监乔·斯洛维克(Joe Slowik)说。“也没有证据表明这些事件具有恶意性质。所以我的首要建议是保持谨慎。”美国官员传递了相同的信息。国务卿马可·鲁比奥(Marco Rubio)将这些事件描述为“完全独立且各不相同”。他说:“我们并未遭受系统性攻击。”国土安全部部长马克韦恩·马林(Markwayne Mullin)同样表示,网络攻击不是原因。

“There is no evidence that they are linked whatsoever,” said Joe Slowik, director of threat research and cyber engineering at the risk intelligence firm Dataminr. “Or that any of these events have been malicious in nature. So my first argument is for caution.”US officials had the same message. Secretary of State Marco Rubio described the incidents as “all separate and distinct.”He said, “We are not under systemic attack.”Homeland Security Secretary Markwayne Mullin similarly advised that a cyberattack wasn’t the cause.

尽管对部分故障的解释细节很少,但一个共同主题似乎是更换旧技术出现了问题。国家情报总监杰伊·克莱顿(Jay Clayton)表示,美国国家安全局(NSA)的中断与修复 PHP 编程语言缺陷的一次更新有关。

Although the explanations offered for some of the outages were scant on details, a common theme appeared to be problems with replacing old technology. Director of National Intelligence Jay Clayton said the disruption at the NSA was related to an update to a flaw in the PHP programming language.

负责监管 PHP 的组织——该技术支撑着无数网站——在 10 月表示,已在其技术中发现两个关键缺陷。这些缺陷已存在约十年,直到网络研究人员利用 AI 模型搜寻漏洞时才被发现。该组织披露了漏洞详情,银行、航空公司和政府纷纷将网站下线,以便在黑客入侵前完成修复。

The organization that oversees PHP, which underpins scores of websites online, said in October that it had discovered two critical flaws in its technology. The flaws had existed for roughly a decade and were only found when cyber researchers used AI models to scour for any bugs. The organization disclosed details of the vulnerabilities, and banks, airlines and governments yanked their websites offline in order to fix them before hackers broke in.

11月18日,在华盛顿国会山举行的众议院国土安全委员会会议上,国土安全部长马克韦恩·马林(左)、国家情报总监杰伊·克莱顿和联邦调查局副局长克里斯托弗·拉亚。Homeland Security Secretary Markwayne Mullin, from left, Director of National Intelligence Jay Clayton, and FBI Deputy Director Christopher Raia during a House Homeland Security Committee on Capitol Hill in Washington on Nov. 18.

那次 PHP 事件成了近期故障以及预计后续将出现故障的一种预警。

That PHP incident turned out to be a kind of warning for the recent outages, and the ones expected to come later.

人类黑客可能需要数周才能找到可利用的软件弱点,而 AI 模型已证明能在数小时内完成大部分同类工作。事实上,Anthropic PBC、OpenAI、Alphabet 旗下 Google 以及其他众多公司的 AI 模型正帮助以空前速度发现创纪录数量的计算机漏洞。根据美国国家漏洞数据库(US National Vulnerabilities Database)的数据,该数据库追踪已知的数字安全漏洞,AI 模型有望在 2026 年发现的网络漏洞数量是 2025 年的两倍。一旦发现漏洞,就需要有人修复它。这才是难点所在。

Whereas human hackers might take weeks to find software weaknesses to exploit, AI models have proven they can do much of the same work within hours. In fact, AI models from Anthropic PBC, OpenAI, Alphabet Inc.’s Google and numerous others are helping find record numbers of computer bugs faster than ever. AI models have been on pace to find twice as many cyber flaws in 2026 as they did in 2025, according to the US National Vulnerabilities Database, which tracks known digital security holes. Once a bug is found, someone needs to squash it. That’s the hard part.

Noma Security安全战略副总裁克里斯·休斯表示,在修复这些漏洞方面存在“人力瓶颈”。在10月发现并披露PHP问题后,安全专家预测,受影响的组织需要数月时间才能升级其系统。当它们需要修复越来越多的漏洞时,这件事会变得更加困难。

Chris Hughes, vice president of security strategy at Noma Security, says that there’s a “human bottleneck” when it comes to fixing the flaws. After the PHP issues were discovered and disclosed in October, security experts predicted that it would take months for affected organizations to upgrade their systems. That becomes much harder when they need to fix more and more flaws.

美国民主党参议员、参议院情报特别委员会副主席马克·华纳表示,新技术发现漏洞的速度前所未有,这给那些并非为应对如此变化速度而构建的系统带来了巨大压力。

Democratic US Senator Mark Warner, vice chairman of the Senate Select Committee on Intelligence, said that the unprecedented speed with which new technology is unearthing vulnerabilities is putting immense pressure on systems that weren’t built to handle that pace of change.

Dataminr的斯洛维克说:“人们不得不用比几年前更少的资源,在更短的时间内做更多的事,这会导致错误。”他说:“团队行动太快,同时处理太多问题。”

“People have to do more in less time, with fewer resources than a few years ago and that leads to errors,” said Slowik, of Dataminr. “Teams are moving too fast and dealing with too many issues simultaneously,” he said.

伦敦大学城市圣乔治网络安全研究所所长拉杰·拉贾拉詹表示,随着AI应用加速,世界可能会看到更多宕机及相关事件。网络安全专家表示,在严重漏洞被发现后的数天和数周内,中断风险会升高,因为各组织会争相保护自己,有时并未测试其修复措施是否会造成新问题。

The world is likely to see more outages and related incidents as AI adoption accelerates, said Raj Rajarajan, director of the Institute for Cybersecurity at City St. George’s, University of London. The risk of disruptions is heightened in the days and weeks after a serious vulnerability is discovered, as organizations rush to protect themselves, sometimes without testing that their fixes don’t create new problems, cybersecurity experts have said.

阅读媒体原文 ↗