CAPTCHA (kapcha)

CAPTCHA — with engl. Completely Automated Public Turing test to tell Computers and Humans Apart — completely automated public Turing test for distinguishing of computers and people. The main idea is in offering the user such problem which is solved with ease by the person, but is extremely difficult and labor-consuming for the computer. Kapcha is developed in Carnegie Mellon University and afterwards received continuation in the project under the name reCAPTCHA. In 2009 the reCAPTCHA system was purchased by Google company.

2018

The fastest and exact algorithm for cracking of CAPTCHA is created

At the end of December, 2018 it became known of development of the fastest and exact algorithm of machine learning which can crack the text CAPTCHA systems.

This new algorithm developed by scientists from Lancaster University (Great Britain), Northwestern University (China) and Peking University (China) is based on the concept of GAN. It is a special class of AI algorithms which are used in cases of unavailability of large volume of information. Huge databases for training usually are required for the classifying algorithms of machine learning, but GAN uses so-called "generative" approach for creation of similar information on the basis of available. These "generated" data then are analyzed by the general algorithm.

At the end of December, 2018 it became known of development of the fastest and exact algorithm of machine learning which can crack the text CAPTCHA systems

Scientists applied this concept to cracking of text CAPTCHA which in the majority of the previous researches were estimated only in terms of classical algorithms of machine learning. Researchers claim that in the real scenario the malefactor will not be able to generate millions of CAPTCHA on the real website or API without detection. Therefore for the research they used only 500 text CAPTCHA from each text service.

The received algorithm could recognize quickly and precisely the text, and such approach was more effective and cheap, than any other available systems. According to the statements of researchers the algorithm is capable to crack any text of CAPTCHA within 0.05 seconds when using the normal PC. It means that malefactors will not need to buy expensive servers of cloud computing and to pay for them to crack text CAPTCHA in real time on websites.

Researchers recommend to owners of websites to implement the alternative measures for detection of bots using several security levels, such as comparison of templates of use, location of the device or biometric data.^[1]

reCAPTCHA v3 exit

At the end of October, 2018 Google submitted the new version of a system of protection of the websites against reCAPTCHA robots. The technology is interesting that it does not demand from the user of special actions, like text typing or the choice of pictures, confirmatory that he is a person, but not the robot.

Within the last ten years of reCAPTCHA constantly improved the technology. If in brief, reCAPTCHA helps to protect your websites without inconveniences for users and gives you more opportunities to decide what to do in risk situations — the manager on product management of Google   Wei Liu noted.

reCAPTCHA analyzes actions of visitors of the website, including the movement of a mouse and interaction with pages, and distinguishes robots from real people.

Google provided CAPTCHA which will recognize bad traffic

The site administrator can set the trigger level at which a system will require from the user to undergo additional testing. For this purpose they add to pages or their  sections of action tag.  The requirement to undergo telephone verification,  in response to  attempt to leave the comment, to get access to  a user profile or to  the history of transactions can be one of options of such actions.

reCAPTCHA v3 also takes into account signals according to several pages, but not on one as it was in the first and second versions of technology. reCAPTCHA v3 is already available to installation on the websites.^[2]

Notes