Captcha Security can be fooled with Artificial Intelligence—Research Claims – Research Snipers

Captcha Security can be fooled with Artificial Intelligence—Research Claims

Computer programmers have developed artificial Intelligence AI that can outsmart the captcha security system of a website.

Captcha Security

Captcha security system is used to keep the bots away from the website in order to curb spam and maintain the website’s security. Captcha challenges people to prove that they are human by recognizing letters, numbers that machines would struggle to do correctly.

Researchers have successfully developed an algorithm that imitates how the human brain responds to these visuals and images. The Neural Network could identify numbers and letters from the images. The research is carried out by an artificial intelligence company Vicarious funded by both CEO of tech companies Jeff Bezos and Mark Zuckerberg, the study is published in Journal “Science”

Google says their captcha is so difficult to pass so that only 87% of humans are able to do it correctly. However, the researchers claim that their algorithm can pick the numbers and letters from the image and solve it easily.

The use of Neural Networks

Computer scientists use neural networks to recognize images; neural networks are a large network of computers trained to solve complex mathematical problems. Inspired by human brain these networks have hundreds of layers, the answers from all the layers are combined eventually to produce the final result which has more accuracy.

The team of researchers at Vicarious developed Recursive Cortical Network (RCN), this is a software which mimics actual processes in the human brain and requires less computing power than a normal neural network.

Human brains are so efficient that they can recognize the objects and patterns in obscure images, the Vicarious team is developing algorithms for RCN aimed at identifying objects by analyzing pixels in the image in order to see if they match and outlines an object.

Captcha Cracking

The company has announced in 2013 that it had successfully cracked text-based Captchas including Google, PayPal, Yahoo and Captcha.com with 90% of accuracy.

But as Google claims, reCaptcha is much difficult to solve, the company says its software can now pass Google’s reCaptcha test with 66.6 percent accuracy.

Simon Edwards a cybersecurity architect commented on the researchers work and said, “Whatever the researchers have developed will become mainstream in 4 to 5 months, Captcha days are numbered, he told BBC.