According to Google, this algorithm was designed to be used to help Google analyze hard-to-read signs and house numbers captured by its Street View cameras. By solving this issue, they will be able to more accurately match images with locations on a map, but in the process it seems that this algorithm could also be used to solve CAPTCHAs.
However it seems that Google will not be putting their algorithm to the test in trying to break other CAPTCHAs. Instead the company will use this knowledge that they have gained to further strengthen reCAPTCHA, Google’s own CAPTCHA service. In a post by Vinay Shet, reCAPTCHA’s produce manager, “Thanks to this research, we know that relying on distorted text alone isn’t enough.”
It’s actually ironic that CAPTCHAs, designed to be read by humans, is more accurately read by an algorithm when humans are the ones who probably make more mistakes when it comes to identify the words/numbers!