NSA Network Outage Enters Second Day as Officials Blame Maintenance Affecting PHP
The US National Security Agency (NSA) spent a second day on Wednesday working to restore its internal communications network, known as NSA Net. The system suffered intermittent outages on Tuesday, and the agency told staff to stagger their arrivals at its Fort Meade headquarters in Maryland. By evening it had closed the campus to most employees, according to sources.
Non-essential personnel received a "Code Gray" notification late Tuesday night telling them to arrive at 10 a.m. Wednesday. That time was later pushed to the afternoon and could change again, a source familiar with the situation said. The same person said the network is working, but leaders are still staggering staff because the login process is the most taxing function on the system. The source guessed full restoration could take another day or two, and insisted classified national security work has not been impeded.
Director of National Intelligence Jay Clayton said the outage was due to maintenance affecting PHP, a common programming language. The source confirmed his account was correct. It remains unknown whether the vulnerability involved was found by AI or by humans. The NSA did not immediately respond to a request for comment.
The incident follows several high-profile technical disruptions in the past week, including a global outage of Microsoft products, a nationwide flight stoppage and problems with bank services. Senate Minority Leader Chuck Schumer called for a dedicated red team to test the assumption that the outages are routine, and asked for findings by December 1. Testifying before the House Homeland Security Committee, Clayton said the timing raises fair questions, but that the incidents are separate service outages tied to maintenance and technical changes, not a coordinated or state-sponsored attack.
Why it matters
Even a highly resourced agency was knocked back by maintenance work touching a common component, which shows that routine change can cause outages on par with an attack. The login process was reported as the most strained part of recovery, so teams should assume that authentication will be the bottleneck when restoring systems. Leaders also face the problem that the public cannot easily tell a maintenance failure from an attack when several outages cluster together, so incident communications need to be ready for that scrutiny.
What to do
- Test maintenance changes to shared components such as PHP in a staging environment before rolling them out.
- Plan staggered staff access so that login demand does not overload systems during recovery.
- Prepare a communications plan that states early whether an outage is maintenance-related or a suspected attack.