Google has figured out how to solve its own CAPTCHAs. In a paper published earlier this week, Google researchers revealed that they have created an algorithm that can solve CAPTCHAs with a 99.8% success rate.
The technology was developed with the intent of improving address accuracy in Google’s Street View, yet had surprising, and even more effective results when applied to the anti-spam puzzle mechanism used on millions of websites. Google can, by comparison, use the technology to identify street signs about 90% of the time.
Google’s own CAPTCHA service, known as reCAPTCHA, seems to have overall benefitted from the news. “Thanks to this research, we know that relying on distorted text alone isn’t enough,” says Vinay Shet, the product manager of reCAPTCHA.
Typically, these anti-spam tools use distorted text images to test users — the text is legible to humans, but not to machines. Google had already updated its CAPTCHA system last October in order to rely less on text recognition and more on the behavior of users themselves. “[This] doesn’t mean that reCAPTCHA itself is broken or ineffective,” says Shet. “These findings have helped us build additional safeguards against bad actors in reCAPTCHA.”
“I think this is a double edged sword,” says Chris Traxler, Managing Partner at TSI. “The ability for Google to be able to do this will make it much more efficient, but at the same time, they’ll retain more analytical information on us and know what we’re doing. That comes at the expense of our privacy.”
The CAPTCHA has seen many transformations over the years since it first appeared, and they are now used more than 300 million times a day by various users. It prevents bots from using various services and collecting sensitive information. Many users have criticized the technology for often creating messages that are inscrutable to humans and machines — and Google’s own ability to solve it better than most people doesn’t seem to be helping. For now, though, it seems that the CAPTCHA is here to stay.