"Captcha (Completely Automated Public Turing test to tell Computers and Humans Apart) challenge-response systems, which are used to prevent accounts being created until a user correctly identifies letters in an image, are designed to ensure requests are made by a human rather than an automated program. The technique has been used to defeat automatic sign-ups to email accounts by services including Yahoo! Mail and Gmail for years, and hackers are increasingly successful in defeating the approach. For example, the HotLan Trojan has created more than 500,000 spam email accounts with Hotmail, Yahoo! and Gmail since its arrival back in July 2007.

Websense reckons the latest Gmail Captcha hack is the most sophisticated it has seen to date. Unlike Live Mail Captcha breaking, which involved just one zombie host doing the entire job, the Gmail breaking process involves two compromised hosts. Each of the two compromised hosts applies a slightly different technique to analyzing Captcha, as explained in a posting by Websense.

Even using the two techniques, only one in every five Captcha-breaking requests are successful. It's a fairly low percentage, but one that's still more than workable in the case of automated attacks."

Article Link: http://www.theregister.co.uk/2008/02/25/gmail_captcha_crack/