AI security fixes crashed banks, flights, and the NSA because old systems couldn't handle the load
Everyday services you rely on — your bank app, airline trips, work email — were disrupted when systems tried to patch AI-found flaws faster than they could cope.

What went down
November has been an unusual month for tech disruptions. A nationwide ground stop grounded all US air travel for hours. Core Microsoft services — including Outlook and Teams, which millions of workers depend on daily — were unreachable for days. Apple's App Store and Apple Pay also went offline. US banks experienced major disruptions. United Airlines had to delay and cancel countless flights. And earlier this week, the NSA's own in-house communications network for classified information went dark 1.
Lawmakers have been left puzzled, trying to piece together what caused such wide-ranging failures across so many different sectors 1.
AI finds flaws, systems buckle under the fixes
According to Wall Street and NSA sources speaking to Congress, the outages were triggered by a wave of system updates designed to plug cybersecurity holes that AI models have been rapidly uncovering. As AI tools get better at spotting weaknesses in software, companies and government agencies are scrambling to patch them — sometimes all at once 1.
"These models are identifying massive vulnerabilities, and that increases the urgency to patch right away," a congressional official told the Washington Sun. "They're trying to push through a lot of patches all at once — more than the system was designed for" 1.
Anthropic's Claude Mythos model, for example, reportedly found more than ten thousand vulnerabilities in existing software earlier this year — illustrating just how fast AI is finding flaws humans may have missed 1.
Not a cyberattack — but close
Department of Homeland Security secretary Markwayne Mullin told reporters there was "98 percent certainty this isn't a cyberattack" 1. He said new technology is being pushed "on top of really old computer systems," which simply could not handle the strain .
But officials also acknowledged the timing is worrying. While these outages were self-inflicted, the same AI tools are being used by hackers to find weaknesses — setting off an accelerating arms race between defenders and attackers 1.
Nathan Calvin, chief strategy officer at Encode, said the situation reveals something unsettling: "the extent to which safety-critical infrastructure we rely on is based on software with large vulnerabilities on legacy systems" 1. He warned the pressure will only grow as AI models become more advanced .
For ordinary users, the takeaway is sobering — the apps and services you reach for every day run on foundations that were never built to withstand this kind of rapid, AI-driven change 1.
This story reports on a trend for information only. It is not advice or encouragement to do any of it.
Corrections and updates
- Corrected on 19 November 2026. An earlier version said AI models like Anthropic's Claude Mythos have found over 10,000 vulnerabilities, implying a broader claim. The corrected version accurately reflects the source, which specifically stated that Anthropic claimed its Claude Mythos model alone had found more than ten thousand vulnerabilities in existing software earlier this year.
Key facts
- Multiple major US services crashed in November, including banks, flights, Microsoft, Apple Pay, and the NSA's classified network
- Outages were caused by rushed AI security patches, not by hackers
- DHS secretary says there is 98% certainty this was not a cyberattack
- - Anthropic claimed its Claude Mythos model had found more than ten thousand vulnerabilities in existing software earlier this year
- Older legacy systems could not handle the volume of urgent security patches
Sources
If this story concerns you or your company, or uses something that belongs to you, ask for a correction or for it to be taken down. How requests work
Keep reading
The weekThis Week in AI New Models Flood In and Agents Get SaferHerald of AI
This Week in AI New Models Flood In and Agents Get Safer
Big open-weight releases, speed breakthroughs, and a push to make AI agents more reliable and secure.
Business and rulesMirendil raises $1 billion to use AI agents for faster model developmentHerald of AI
Mirendil raises $1 billion to use AI agents for faster model development
A new startup founded by former Google DeepMind and Anthropic researchers says its automated platform lets small teams build large AI models quickly and cheaply.
Apps and productsJ.Crew launches an AI stylist powered by Anthropic's Claude modelsHerald of AI
J.Crew launches an AI stylist powered by Anthropic's Claude models
Shoppers in the US can now describe an occasion and get a full outfit from J.Crew — all using Claude instead of hunting for each piece one by one.