US Outage Wave: Officials See No Attack, Experts Blame Rush to Patch After AI Discoveries
In one week in the US, Microsoft's email service went down, flights were grounded, and banks, Apple and a classified NSA network all had outages. Secretary of State Marco Rubio said the evidence that these outages are connected to each other or to a hacking attack is "zero". Officials and experts offer a simpler explanation: companies are rushing to update their systems, and AI is finding vulnerabilities faster than they can be closed. The takeaway for engineers: prepare urgent patches and a rollback plan in advance.
AI-processed from Guardian; edited by Hamidun News
In a single week, several major US systems failed at once: Microsoft email, air traffic services, online banking, Apple apps and even the NSA's classified network. The chain looks like a coordinated attack, and that is exactly what reporters asked the Secretary of State. Marco Rubio's answer was blunt: the US has "zero" evidence that the outages are connected to each other or to the actions of hackers, criminals or another state. In his words, the country is "not under a systemic attack".
For developers and product owners, a different point in this story matters more: officials and independent experts agree that the cause lies in the updates themselves, and that artificial intelligence is pushing companies to make them.
What exactly broke
The series began last Thursday, when Microsoft's email and other workplace tools stopped working, affecting millions of employees. The company blamed a "previously performed network capacity increase". Flights across the country were then halted in the longest "ground stop" since 9/11, and the Federal Aviation Administration (FAA) cited a router update.
On Monday the problems moved to banks. Customers of several large financial institutions and of Zelle could not log in to their online accounts after an outage at JPMorgan Chase. The bank called it a "software error" and stressed that it had nothing to do with a cyberattack. The same day, United Airlines fixed an internal glitch that had been delaying departures. On Tuesday the NSA took its internal network for transmitting classified data offline for "scheduled maintenance", and Apple confirmed and fixed problems with Apple Pay and iCloud.
Director of National Intelligence Jay Clayton explained it this way: everyone understands that systems are overdue for updates, but the updates drag on or get rolled back differently than planned. The hardest-hit spots are those where one system hands work off to another. Democratic Senator Mark Warner agreed: so far everything points to legitimate attempts to strengthen defenses, not to malicious intent.
Where AI comes in
In parallel, researchers at the US company Calif reported a vulnerability in Java, which is used by 90% of Fortune 500 companies. The flaw allows malicious code to be run remotely on someone else's system. It was found with OpenAI technology. The same team recently used AI to discover bugs in PHP.
Professor Steven Murdoch of University College London clarified the scale: the problem does not affect every Java program. It sits in the Base64 library that ships with one of the Java versions. It is serious, but it is not a case of "every Java is broken". The UK's National Cyber Security Centre (NCSC) urged organizations to update, and authorities in the US and South Korea asked Java users to install the updates.
Security expert Tal Beri does not think the outages are directly connected, but he sees a common theme: AI. In his assessment, IT infrastructure is fragile, and AI puts pressure on it twice: through the rebuilding of systems for neural networks, and through the urgent patches that AI-driven vulnerability discovery demands. He notes that the volume of findings has grown since April, when Anthropic released its Mythos model with a striking ability to find holes in software, and OpenAI also released models for finding unknown bugs.
What follows from this
The accounts do not fully agree. Cynthia Kaiser, a former deputy head of the FBI's cyber division, called it "plausible" that warnings about the plans of specific cyber adversaries spurred the race to update. She is more concerned about scammers who exploit the confusion to steal money from companies and ordinary people. Donald Trump, when asked about the outages, brushed them off: "That's life. When you're big and powerful, this happens."
An important caveat: there is no public evidence for either version. The statements by Rubio, Clayton and Warner are officials' assessments, not a technical analysis, and the companies cite different causes: a network update, a router, a software error, scheduled maintenance.
The practical takeaway for teams is this. If AI tools find vulnerabilities faster than before, patches will arrive more often and with more urgency. That calls for a dependency inventory (for example, which Java version and which bundled libraries you run in production), staged rollouts with the ability to roll back quickly, and a review of the places where one system hands data to another. Finally, expect phishing during high-profile outages: emails about an "urgent update" are especially dangerous right now.
- Nov 20, 2026Nvidia's stakes in startups grew from $22 billion to $65.5 billion, but without Huang the deals are in question
- Nov 20, 2026MacRun rents out Mac minis for AI agents at $169 a month and earns ~$2K
- Nov 20, 2026Anthropic goes public: shares open at $111.50 versus an $83 offer price, up 34%
- Nov 20, 2026Orange fine-tunes Qwen3 for Wolof: five open models from 0.6B to 14B parameters
Key Questions Answered
What exactly broke?
The series began last Thursday, when Microsoft's email and other workplace tools stopped working, affecting millions of employees. The company blamed a "previously performed network capacity increase". Flights across the country were then halted in the longest "ground stop" since 9/11, and the Federal Aviation Administration (FAA) cited a router update. On Monday the problems moved to banks.
Where AI comes in?
In parallel, researchers at the US company Calif reported a vulnerability in Java, which is used by 90% of Fortune 500 companies. The flaw allows malicious code to be run remotely on someone else's system. It was found with OpenAI technology. The same team recently used AI to discover bugs in PHP. Professor Steven Murdoch of University College London clarified the scale: the problem does not affect every Java program. It sits in the Base64 library that ships with one of the Java versions.
What follows from this?
The accounts do not fully agree. Cynthia Kaiser, a former deputy head of the FBI's cyber division, called it "plausible" that warnings about the plans of specific cyber adversaries spurred the race to update. She is more concerned about scammers who exploit the confusion to steal money from companies and ordinary people. Donald Trump, when asked about the outages, brushed them off: "That's life. When you're big and powerful, this happens." An important caveat: there is no public evidence for either version.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.