I only partly agree with Bruce Schneier when he criticizes the anti-virus industry for not finding the Sony rootkit for all the months before it was discovered by Sysinternals Mark Russinovich.
Most of Schneiers criticism actually relates to the sloth they exhibited in adding detection once the threat was exposed, and the fact that they tended just to remove the rootkit, not the unstable DRM components.
There are numerous difficult questions in this scenario. Its easy and cheap to lash out at various industries for not doing better for consumers, but many of the things Schneier and his blog comment writers yell about just couldnt be helped.
The fact remains, and will continue to remain after this, that its not always easy to determine what is a threat and what is legitimate system software performing privileged operations. For instance, FORMAT.EXE can destroy your hard disk, but its not malware. How should an anti-virus program know the difference? The more important fact is that in this case two companies that should be trustworthy, Sony and First 4 Internet, betrayed that trust. This is enough to make a mess of any reasonable security system.
Incidentally, I may as well point out here that a great way to protect yourself against the First 4 Internet rootkit and many other threats is to disable Autoplay, the Windows feature that automatically runs software when a mass storage volume is attached to the system. Its most famous for CDs, but works for all media—USB keys, for example. Click here to read how to disable it for all drives on the system.
I was recently on a security discussion group with the great Dr. Solomon (author of a famous anti-virus product later bought by McAfee). The good doctor posed the (rhetorical, of course) question of why the Sony rootkit is something that an anti-virus product should detect. The main answer, and the reason its a threat, is not that it performs DRM functions but that it hides itself sloppily, opening up the system for exploitation by third parties. (Dr. Solomon still has lots of friends, but no financial interest, in the anti-virus industry.)
If this is true, the doctor continues, surely an old, unpatched copy of Internet Explorer qualifies as a threat to be flagged by anti-virus software. After all, it has vulnerabilities that are being actively exploited in the wild.
What should we expect
The biggest part of this problem, specifically the problem of how anti-virus programs should decide what to block, is more one of policy than anything technical. Dr. Solomons question is very reasonable, and yet of course its reasonable for an anti-virus program to flag and remove the Sony rootkit, but unreasonable for it to remove unpatched copies of IE.
It is reasonable for security software to look for unpatched vulnerabilities and flag them for the user. Trends PC-cillin Internet Security Suite, which I run on one of my computers, looks for unpatched Windows vulnerabilities by default once a week. This is a useful service, and I have another service available, from Microsoft, to remediate these problems. I dont want Trend removing these programs.
The bottom line is that users intend for Internet Explorer to be on their computers (or at least they know its supposed to be there), while malware programs, and the Sony rootkit, are there through subterfuge. This alone is a good reason to remove them.
I am not as upset as Bruce Schneier at the AV companies for missing the rootkit, although they should have moved faster once its existence was revealed. As he says, only F-Secure moved quickly, and thats because they are at the forefront (with Mark Russinovich and Sysinternals) of rootkit detection under Windows. But prior to that there would be no reason to expect them to find it.
And does anyone really expect that anti-virus companies should have to acquire every version of every major commercial program to test it for malware? Better still, should they have to test all the music CDs out there? Its just not reasonable.
There is a separate issue of heuristics. If it was reasonable for anti-virus products not to have a signature for the rootkit before it was exposed, was it reasonable for their heuristics to miss it? The heuristics in most of these products are pretty lame, but there are some that do a good job of detecting malware generally. Should these have missed it? Did they? Turns out we still dont know if any of these products detected it, but were looking into it.
As pointed out in the blog and comments attached to it, companies like First 4 Internet actually work with anti-virus companies to see that their programs are not detected as malicious. This is a good thing. It would be bad for everyone if anti-virus programs were to start falsely detecting legitimate software as malicious.
(I should point out that a report on News.com originally stated that First 4 Internet worked with Symantec specifically, implying that they did so on the rootkit. This was later corrected to say that the two companies have worked on First 4 Internets digital imaging software.)
This is yet another reason why government, the industry and lawyers need to make an example of Sonys behavior in this affair. They betrayed a very serious trust by installing on their customers systems software that any normal person would consider deserving of detection by anti-virus software. Legitimate software companies beware: Sony has undermined a system that helps your products run smoothly on your customers computers. Are you going to let them get away with it?
Security Center Editor Larry Seltzer has worked in and written about the computer industry since 1983.
More from Larry Seltzer