Yahoo's CAPTCHA Security Reportedly Broken

Yahoo may soon see a surge in spam coming from Yahoo Mail accounts.

"John Wane," who identifies himself as a Russian security researcher, has posted software that he claims can defeat the CAPTCHA system Yahoo uses to prevent automated registration of free Yahoo Mail accounts.

CAPTCHA stands for Completely Automated Public Turing test to tell Computers and Humans Apart. It's a technique that presents an image depicting distorted text that people, but not machines, can identify.

Large e-mail service providers like Google, Microsoft, and Yahoo present CAPTCHA images to users signing up for new accounts to make sure that there's a real person behind the registration information. These companies do so to discourage spammers from using automated methods to register thousands of free online accounts to send spam.

CAPTCHAs are also used to prevent spam in blogs and other online forums, automated ballot stuffing for online polls, and automated password guessing attacks.

"Few months ago, we received information that [a] Yahoo CAPTCHA recognition system exists in the wild with the recognition rate about 30%," Wane says in a blog post. "So we decided to conduct few experiments. We explored Yahoo CAPTCHA and designed a similar system with even better recognition rate (about 35%)."

Various automated methods exist to defeat CAPTCHA schemes but the CAPTCHAs used by Google, Microsoft, and Yahoo have remained difficult for computers to crack.

If the software works as advertised, and it's not clear that it does, it could force Yahoo and other companies to spend yet more money to defend against spammers.

"We are aware of attempts being made toward automated solutions for CAPTCHA images and continue to work on improvements as well as other defenses," a Yahoo spokesperson said in an e-mailed statement.

John Orbeton, strategic product manager for IronPort, said that if the software works, "it could be used for spam. It could be used for phishing. It depends on the motivation of the attacker." The claimed rate of success, 35%, he said, "could create a fairly significant number of e-mail accounts."

It is ironic, Orbeton added, that image-recognition technology, which is being used to defend against the current generation of image spam, should be used by spammers to create more spam.

Not that there's any shortage of the stuff. "In 2007 we saw spam volumes increase 100%," Orbeton said. "That comes out to around 20 spam messages per day for everyone on the planet, whether they have e-mail or not."