It is still out there.
Like a ghost ship, a rogue software program that glided onto the Internet last November has confounded the efforts of top security experts to eradicate it and trace its origins and purpose, exposing serious weaknesses in the world's digital infrastructure.
The program, known as Conficker, uses flaws in Windows software to
Alarmed by the program's quick spread after its debut in November, computer security experts from industry, academia and government joined forces in a highly unusual collaboration. They decoded the program and developed antivirus software that erased it from millions of the computers. But
"It's using the best current practices and state of the art to communicate and to protect itself," Rodney Joffe, director of the Conficker Working Group, said of the malicious program. "We have not found the trick to take control back from the malware in any way."
Researchers speculate that the computer could be employed to generate vast amounts of spam; it could steal information like passwords and log-ins by capturing keystrokes on infected computers; it could deliver fake antivirus warnings to trick naive users into believing their computers are infected and persuading them to pay by credit card to have the infection removed.
There is also a different possibility that concerns the researchers: That the program was not designed by a criminal gang, but instead by an intelligence agency or the military of some country to monitor or disable an enemy's computers. Networks of infected computers, or botnets, were used widely as weapons in conflicts in Estonia in 2007 and in Georgia last year, and in more recent attacks against South Korean and United States government agencies. Recent attacks that temporarily crippled Twitter and Facebook were believed to have had political overtones.
Yet for the most part Conficker has done little more than to extend its reach to more and more computers. Though there had been speculation that the computer might be activated to do something malicious on April 1, the date passed without incident, and some security experts wonder if the program has been abandoned.
The experts have only tiny clues about the location of the program's authors. The first version included software that stopped the program if it infected a machine with a Ukrainian language keyboard. There may have been two initial infections--in Buenos Aires and in Kiev.
Wherever the authors are, the experts say, they are clearly professionals using the most advanced technology available. The program is protected by internal defense mechanisms that make it hard to erase, and even kills or hides from programs designed to look for botnets.
A member of the security team said that the Federal Bureau of Investigation had suspects, but was moving slowly because it needed to build a relationship with "noncorrupt" law enforcement agencies in the countries where the suspects are located.
An FBI representative in Washington declined to comment, saying that the Conficker investigation was an open case.
Eventually, university researchers and law enforcement officials joined forces with computer experts at more than two dozen Internet, software and computer security firms.
The group won some battles, but lost others. The Conficker authors kept distributing new, more intricate versions of the program, at one point using code that had been devised in academia only months before. At another point, a single technical slip by the working group allowed the program's authors to convert a huge number of the infected machines to an advanced peer-to-peer communications scheme that the industry group has not been able to defeat. Where before all the infected computers would have to phone home to a single source for instructions, the authors could now use any infected computer to instruct all the others.
In early April, Patrick Peterson, a research fellow at Cisco Systems in San Jose, Calif., gained some intelligence about the authors' interests. He studies nasty computer programs by keeping a set of quarantined computers that capture and observe them--his "digital zoo."
He discovered that the Conficker authors had begun distributing software that tricks Internet users into buying fake antivirus software with their credit cards. "We turned off the lights in the zoo one day and came back the next day," Peterson said, noting that in the "cage" reserved for Conficker, the infection had been joined by a program distributing an antivirus software scam.
It was the most recent sign of life from the program, and its silence has set off a debate among computer security experts. Some researchers think Conficker is an empty shell, or that the authors of the program were scared away in the spring. Others argue that they are simply biding their time.
If the misbegotten computer were reactivated, it would not have the problem-solving ability of supercomputers used to design nuclear weapons or simulate climate change. But because it has commandeered so many machines, it could draw on an amount of computing power greater than that from any single computing facility run by governments or Google. It is a dark reflection of the "cloud computing" sweeping the commercial Internet, in which data is stored on the Internet rather than on a personal computer.
The industry group continues to try to find ways to kill Conficker, meeting as recently as Tuesday. Joffe said he, for one, was not prepared to declare victory. But he said that the group's work proved that government and private industry could cooperate to counter cyberthreats.
"Even if we lose against Conficker," he said, "there are things we've learned that will benefit us in the future."
Entire contents, Copyright © 2009 The New York Times. All rights reserved.