McAfee messes up Google China attack analysis
Researchers link unrelated files to Aurora attack
By Robert McMillan | Published: 11:45, 31 March 2010
A misstep by McAfee security researchers apparently helped confuse the security research community about the hackers who targeted Google and many other major corporations in cyber attacks last year. McAfee disclosed that its initial report on the attacks, branded Operation Aurora by McAfee, had mistakenly linked several files to the attacks, files that had nothing to do with Aurora after all.
Aurora is a sophisticated spying operation, set up to siphon intellectual property out of major corporations. It has been linked to attacks on Google, Intel, Symantec, Adobe, and other companies. Google took the attacks seriously. Last week it pulled its search engine out of China, in part because of the incident. The files mistakenly linked to Aurora in McAfee's initial research are actually connected to a still-active botnet network of hacked computers that was created to shut down Vietnamese activists.
McAfee investigated more than a dozen companies that had been hit by Aurora and found the Vietnamese botnet on four of these networks, said Dmitri Alperovitch, McAfee's vice president of threat research. At first, McAfee though they were part of the Aurora attack. "It took us a little while to realize that they weren't related," he said.
Related Articles on Techworld
McAfee included four filenames in its original Aurora research that it now says are associated with the Vietnamese botnet: jucheck.exe, zf32.dll, AdobeUpdateManager.exe and msconfig32.sys. McAfee has now "come to believe that this malware is unrelated to Aurora and uses a different set of command and control servers," McAfee Chief Technology Officer George Kurtz said in a blog posting.
Other companies that followed up on McAfee's research were apparently confused too, according to McAfee's Alperovitch. "Some of the other companies that published their analysis on Aurora were analysing this event and just didn't realise it," he said.
One such company was Damballa, Alperovitch said. Earlier this month, Damballa concluded that the Aurora attacks were the work of somewhat amateur botnet writers. That conclusion was disputed by McAfee and other researchers who had been studying the attacks. They were seeing targeted attacks that compromised victims after careful reconnaissance and then used sophisticated techniques to move around the network and quietly move intellectual property overseas.
This type of attack is what computer forensics company Mandiant calls an advanced persistent threat. In it's report, Damballa described it as the work of a "fast-learning but nevertheless amateur criminal botnet team."
"The advanced persistent threat is not a botnet," said Rob Lee, a Mandiant director. Researchers at his company didn't understand how anyone could conclude that these attacks came from amateur botnet writers, he added. But if Damballa's research were based on bad information, that might explain things, he said. "It shows how challenging this job is," he said.
Damballa said it would have a comment on the matter at a later date. In a blog post published a few hours before McAfee's, Google engineer Neel Mehta said that the less sophisticated Vietnamese malware was not related to the attacks on his company.
"Damballa does not have first hand knowledge of our investigation of the attacks we announced in January," a Google spokesman said. "There did seem to be confusion about the two issues on the part of some people, and we've said clearly in our blog post that they were separate."