A string of technology outages has hit some of the biggest names in the US in under a week. Microsoft services went down around the world from 12 to 14 November. On 13 November, the US Federal Aviation Administration stopped all flights nationwide for nearly 5 hours after a communications failure, only the third nationwide ground stop in its history.
JPMorgan Chase customers could not complete some transactions on 16 November, the same day United Airlines delayed departures. On 17 November, Apple services, including the App Store and Apple Pay, went down, and so did the internal classified network of the National Security Agency, the US electronic intelligence agency.
Why? Not cyberattacks. It’s the companies rushing out patches for apparent security flaws their AIs spotted.
Insights on the US tech outages
Officials say they see no sign of an attack. Director of National Intelligence Jay Clayton said there was no evidence of a cyberattack and blamed routine security patching, and Secretary of State Marco Rubio said there was "zero evidence" the outages were connected.
The companies have given their own causes: Apple cited an internal update, the NSA routine maintenance, and JPMorgan a software bug. The Washington Sun, citing 3 unnamed sources, reported that bank executives and the NSA told members of Congress the outages came from attempts to fix security gaps that AI models had found.
"They're trying to push through a lot of patches all at once, more than the system was designed for," one source told the paper. In other words, AI did not attack these systems, but the flood of fixes for weaknesses it uncovered may have overloaded them.
The week also brought a separate warning. On 16 November, Google's Threat Intelligence Group said "the era of autonomous attacks has arrived", reporting that criminals now use AI to study newly released security fixes, work out the flaw each one closes, and build an attack, "all without human involvement". It expects fully autonomous attacks within about 6 months.
Google and the affected companies have not linked that warning to the outages, but Downdetector data shows last week had the most outages of any week in the past year.
What others are saying about the US tech outages
The Washington Sun's Jeff Stein reported the private briefings to Congress. CNBC covered the outages and Clayton's comments, as lawmakers from both parties demanded briefings. Axios noted that to the extent AI is involved, the outages may have more to do with rushing to fix old code than with hacking. BleepingComputer detailed Google's threat report.
Can the world keep up?
The most plausible story so far is not AI attacking the world's systems, but AI finding weaknesses faster than people can safely fix them. That creates a squeeze. Attackers can now use AI to turn a newly published fix into an attack within hours, so companies feel pressure to install fixes immediately.
But rushing large numbers of changes into complex, decades-old systems is exactly how things break. Move too slowly, and you get hacked; move too fast, and you cause your own outage. We saw the same pressure in miniature during the PHP2Shell crisis, and when AI models breached real companies this year.
For South African businesses, the lesson is practical: know which systems matter most, test fixes before rolling them out everywhere, roll them out in stages, and have a way to undo a bad change quickly. Most of all, have a plan for running without your critical systems for a few hours, because this week showed that even the biggest organisations in the world cannot always avoid it.
You might also like our piece on AI agent swarms in Minecraft, Jensen Huang's resignation, and the Ampath data breach.
Get more SA tech and business news and subscribe to The Open Letter.


