This article is more than 1 year old
NSA proposes backdoor detection centre
Will mend holes in software applications
Declaring hidden malware to be "a growing threat", the National Security Agency's cybersecurity chief is calling on the US Congress to fund a new National Software Assurance Center dedicated to developing advanced techniques for detecting backdoors and logic bombs in large software applications.
In prepared testimony before the House Select Committee on Homeland Security's cybersecurity subcommittee last month, NSA information assurance director Daniel Wolf bemoaned an absence of tools capable of scouring program source code and executables for evidence of tampering. "Beyond the matter of simply eliminating coding errors, this capability must find malicious software routines that are designed to morph and burrow into critical applications in an attempt to hide," said Wolf.
The proposed solution: a federally funded think-tank that would include representatives from academia, industry, government, national laboratories and "the national security community," said Wolf, "all working together and sharing techniques."
While accidental security holes dominate the work-a-day security world, government spooks periodically fret over more exotic danger of corrupt software engineers, saboteurs and spies slipping malicious code into commercial software applications used in critical infrastructures and sensitive governmental functions.
In 1999, then-FBI cybercop Michael Vatis warned that cyberterrorists posing as law-abiding programmers could be planting logic bombs in U.S. software while performing Y2K remediation - a theory that never panned out. More recently, U.S. programmers have raised similar security concerns over American companies outsourcing programming work to India, China and other countries.
Cybersecurity thinkers express reserved support for Wolf's proposed national center.
"It's not a bad idea," says John Pescatore, research director for Internet security at Gartner. "It would not take a lot of funding to do. I think the more complicated issue is what do they do with the information. Are they just providing it to the vendors of that software, do they make it public?"
Peter Neumann, a computer scientist at SRI International, points out that researchers have been working on developing automated code analysis tools for decades. "There's a lot of progress in analyzing source code," says Neumann. "But remember that it's still a long way from actually determining that the code you're looking at is the code that's actually in the system at the time."
Neumann says the emphasis on finding malicious code is odd, given the number of security bugs released with commercial software. "The bigger problem is that the software development process is broken, by and large, and systems are being released with literally hundred of flaws," he says. "So trying to find a Trojan horse in one of these systems is only a very small piece of the puzzle. But I think this is an interesting step in a very useful direction."
Indeed, confirmed cases of source code being modified maliciously are few and far between. But Pescatore says the NSA may have some institutional insight into the threat that outsiders lack. "We know they look at lots of software products that other countries are using, to look for vulnerabilities that they can exploit," says Pescatore. "As an intelligence agency, that's part of their job."
In his testimony, Wolf also said that untrustworthy hardware poses a similar threat. "Most microelectronics fabrication in the USA is rapidly moving offshore," said Wolf. "NSA is working on a Trusted Microelectronics Capability to ensure that state-of-the-art hardware devices will always be available for our most critical systems." ®