Computer scientists have created an AI algorithm that works on the same principles of the human eye, and that can break various CAPTCHA systems with accuracies of over 50%.
More specifically, this new system solved Google reCAPTCHAs with 66.6% accuracy, BotDetect with 64.4%, Yahoo with 57.4%, and PayPal image challenges with 57.1%.
For the record, any CAPTCHA system that allows automated systems to break it with an accuracy of over 1% is considered broken.
The 12-man research crew designed their AI algorithm to go through the same steps a human eye and brain go through when viewing an image.
There are algorithm components that recognize the edges of shapes, a component that categorizes the shape, one that takes into account the angle at which an observer is looking at the shape, and then a component attempts to match the shape with a standard form of a letter or number (usually stored in the AI as a Georgia font character).
Researchers named their method a Recursive Cortical Network (RCN), and say it's different from similar AI-based CAPTCHA breakers that rely on a Convolutional Neural Network (CNN) model.
Authors claim their RCN system is superior to CNN systems because it needs far less training and can work outside of the strict rules used to train the algorithm, allowing it to adapt to new CAPTCHA systems.
As a comparison, researchers said their RCN system needed only a few thousand training images, compared to a similar CNN system that needed around 2.3 million.
The CNN system achieved 89.9% accuracy in breaking reCAPTCHA imagery, but researchers said that a slight change in character spacing would throw the CNN system off to 38.4% accuracy, while the same change "results in an improvement in the recognition accuracy" for their RCN system.
Researchers are currently exploring the possibility of deploying their RCN-based adaptive AI bot to parsing images that also contain objects, not only text. If successful, the AI bot could evolve from a CAPTCHA breaker and OCR system to object or facial recognition territory.
More info on this research has been published in this month's Science magazine. The researcher paper is entitled "A generative vision model that trains with high data efficiency and breaks text-based CAPTCHAs " and you can grab a PDF copy here.