April 05, 2011, 2:47 PM — A number of hack attacks recently have made many question the fundamental security of the Internet--hack attacks that have brought into question a system that until now was considered be bullet-proof. However, with appropriate good timing, two new security schemes are coming to the rescue.
HTTPS Hacks: A Brief Synopsis
Hypertext Transfer Protocol Secure--or HTTPS--is the technical name for the padlock system used within Web browsers that shows if a secure connection is in use. It's typically used by online banking sites and Webmail providers, and it relies on a document known as a security certificate, which is issued by a trusted number of certificate authorities (CAs) around the world. Web browsers use these certificates to verify the authenticity of various sites.
However, in March a hacker (or hacker group) called Ich Sun accessed the computer systems for Comodo--the second largest CA in the world--and used its systems to issue fraudulent certificates for Google, Yahoo, Skype, and Hotmail, amongst others. These certificates could be used to make a fake site look legitimate. The certificates were hastily revoked once the hack was discovered, and Microsoft issued an update to ensure that Windows users weren't duped.
A few days ago, Ich Sun hit the headlines again, this time claiming to have breached several more CA systems. It's not clear if Ich Sun issued any other certificates at this time.
This sort of certificate theft isn't a huge threat unless it's used as part of a highly sophisticated hack attack involving taking control of Internet domain-name servers. Feasibly, Ich Sun could have issued certificates for domains that look like the real deal--paypall.com rather than paypal.com, for example. These could then have been used in phishing attacks in which people, seeing the trusted padlock symbol provided by the fraudulent certificate, simply wouldn't be see they were being fooled.
But help is on the way.
New Security for New Threats
The first new development is DNSSEC, as I explained in my earlier story. Assuming this takes off over the coming years (it was only enabled for the .com domain last Thursday), it should provide a reasonable method of proving that we're connected to the site our browser says we are.
Secondly, Google has begun building what it calls the Google Certificate Catalog. This is a Web-accessible database of what Google considers to be valid security certificates. It's updated as frequently as Google's search catalog because the same Web crawler bots collect the data.
Although it's at an early stage right now, the catalog indicates not only if a certificate should be considered valid but also for how long Google has known about it. The simple concept is that, if the Google Certificate Catalog doesn't know about a certificate, it should be considered questionable.
It's possible to probe the database right now but it's not easy and requires a Linux or Mac command-line (the database is stored in the form of a domain-name server so can be queried easily). In future there's a chance the feature will be built-into browsers like Google Chrome or Mozilla Firefox although it will have to be a user-selected option because the results will require interpretation--a certificate that's only been in the database for one day doesn't necessarily indicate shenanigans, for example. It could be that the certificate has recently been renewed.
The project is very similar to Perspectives, an open system created by a handful of security researchers. However, Google's system has the simple advantage of being created by an Internet heavyweight. Google says the catalog is available to whoever wishes to use it.
