Poor Skype. They started out last week with the best of intentions, releasing what they called an independent security evaluation of their VOIP product, and ended up with egg on their virtual faces as high risk security vulnerabilities came to light.
Skype, based in Luxembourg, has positioned its VOIP product as superior to any one elses in the field because the voice data is encrypted. Since Skype hasnt made its encryption scheme public, this has led to some questions on just how secure it is (and how much of a Calera backdoor was built in.) The author of the report, Tom Berson of Anagram Labs, is well respected in the security field and would seem to be a good choice to author such a reassuring effort.
Bersons report was just short of effusive about Skypes security, citing their use of cryptographic primitives like “the AES block cipher, the RSA public-key cryptosystem, the ISO 9796-2 signature padding scheme, the SHA-1 hash function, and the RC4 stream cipher. I looked at the Skype implementation of each of these, and verified that each implementation conforms to its standard and interoperates with reference implementations.”
But upon further examination, there were lots of specifics to quibble about. For instance, the RSA crypto in Skype uses a 1024-bit key. There has been much recent work in the crypto field on the relative ease of breaking such a key. This does not engender trust.
The use of RC4 is also open to question, since if not done properly (like discarding the early and correlatable keystreams) it can be easily defeated.
Further, the Skype key acts as a trusted third party, so Skype itself can impersonate anyone should it want to do so.
But the big unanswered questions on most peoples minds were: (1) what are the techniques that Skype uses to limit the effect on the overall network if a client is compromised? And (2) in a closed source system, how can things be verified; rather than just trust someones opinions, even if that someone seems competent?
Now, it may well be that these questions can be answered by Skype, but the company chose not to do so. However, the overall effect of the paper (at least in the security community) was to raise as many questions as it answered, which I doubt is what Skype had hoped for.
Of course, to make matters worse, vulnerabilities in the code showed up at the same time as the reports release. Skype says that the vulnerabilities affect Skype software for Windows, Mac OS X, Linux and Pocket PC. Skype goes on to say, “Skype can be made to execute arbitrary code through a buffer overflow when Skype is called upon to handle malformed URLs that are in Skype-specific URI types callto:// and skype://.” Also, Skype could launch malicious code “during importation of a VCARD that is in a specific non-standard format.”