The National Security Agency’s classified internal network went down on Tuesday after the agency patched major software flaws that researchers found with AI tools, Director of National Intelligence Jay Clayton said. The agency called it temporary degradation, and said no attack was involved.
What Went Down
The network, called NSANet, lets NSA employees share classified information inside the agency. It went dark on Tuesday, and the agency closed its Fort Meade campus to most staff that night, keeping only essential personnel on site while restoration continued.
A staff voicemail described the outage as code yellow, the second-highest alert level, and told non-emergency staff to arrive in staggered shifts. The alert later moved to green, then to orange, and the campus stayed closed to most staff until further notice.
The NSA said mission-critical capabilities stayed fully operational, and that its infrastructure has layers of redundancy for planned maintenance and unexpected events. A source said restoration was ongoing but not yet complete, and intelligence sharing with allies was not affected.
The Patch That Broke It
Clayton said on Fox Business that the outage followed the patching of major software vulnerabilities discovered last month. “The patching that has gone on following that has caused some of these systems to not come back up when you patch them as expected,” he said.
The flaws sit in PHP, a programming language that powers about 70 percent of the world’s websites. AI tools helped researchers exploit the flaws, setting off a scramble by governments and companies to patch them, and the NSA itself had urged organisations to apply the fixes.
Left unpatched, the flaws could let hackers break into websites and read memory holding sensitive information, including passwords. Clayton said there is no sign that any of the recent tech outages, including the NSA’s, were the result of a cyberattack.
The Wider Claim Stays Unproven
A separate report said bank executives and the NSA told members of Congress that the week’s outages stemmed from attempts to patch flaws found by AI. That report rests on three anonymous sources, and Reuters said it could not verify the claim.
Secretary of State Marco Rubio called the outages “completely unrelated” and said there is “zero evidence” they were caused by malign actors. The affected companies gave their own causes, from a Chase software bug on Monday to an Apple internal update on Tuesday.
Our Read
AI models now find serious flaws faster than institutions can safely fix them, and the NSA just proved the point on its own network. The patch-now doctrine assumes fixes are routine; this week showed that at scale, patching itself can be the outage.
The same week, researchers used AI to find a critical Java flaw in about an hour, for under $200. Discovery is now cheap and fast; safe deployment is not, and the gap between the two is where the next outages will come from.
Why It Matters
The agency that defends American networks could not keep its own network running while it patched. The same week saw the hijack of Microsoft’s official Agent Framework package, so the pressure on defenders is coming from both the flaws and the fixes.
Background
This is not the NSA’s first long outage: a tech failure in 2000 downed its computers for nearly three days. The week also brought a nationwide flight ground stop after a router failure, a Microsoft outage, and a widening breach at Fireworks AI.
Source: CNN report and Reuters report


0 Comments