Computer Glitch Rattles U.S. Intelligence Network

From Monday evening through early Friday last week, the main computers of the National Security Agency failed, causing an unprecedented blackout of information at Fort Meade, Md., where intelligence intercepted around the world is processed, officials said over the weekend.

In what NSA said in a statement was a “serious computer problem,” analytical reports from Fort Meade that turn intercepted foreign telephone, cable and radio messages into meaningful data for the government were halted for 72 hours, starting at 7 p.m. Monday. “Other NSA analysis kept flowing from other parts of the world,” a senior intelligence official said, “but this was not a trivial” failure.

The computer shutdown was caused by a “system overload,” one source said, and was not the result of a Y2K problem, sabotage or hackers invading the system. Another official, who described it as a “software anomaly,” put knowledge of the cause more cautiously.

“As of now,” he said, “there is no evidence other than this was a system stressed to meet day-to-day operational pressures.

“There was a significant loss of processing, but collection continued unaffected. We may have lost timeliness, but we have not lost intelligence.”

The “backlog of intelligence processing is almost complete, and NSA is confident that no significant intelligence information has been lost,” the agency said in a statement.

Almost immediately after a signals intelligence officer saw Monday night that the system had crashed, he turned to other parts of the NSA worldwide system to pick up the processing responsibility, officials said.

To keep up on key early-warning issues during the failure, sources said the U.S. intelligence community turned to other NSA intercept assets in the hands of the CIA and the military. In addition, the NSA regularly exchanges information with allied intelligence agencies.

Early Friday, after calling in various contractors and having personnel work around the clock, fixes had “brought the operation back to operational stability,” the senior official said. As of Saturday, processing had largely been restored to 90% to 95% of operational capability.