A Novel Approach using both Encryption and Graphical Passwords for Artificial Intelligence Security Threats

Authors: Rohit Vibhandik, Nisha Bhati

Certificate: View Certificate

Abstract

Most security based systems of primitives rely on the use of complex mathematical problems to be solved by attackers using either brute force of back engineering to decipher confidential information or credentials. The success of such algorithms lies in the fact that how quickly the attacking computation outruns the rate at which the complexity of the algorithm grows.[1] In this paper, a novel multiple encryption security primitive is proposed which is based on captcha for securing confidential paradigms which uses a dynamic modelling of the number of captcha images as well as the number of co-ordinates for securing entries or log-ins. It has been shown that such a mechanism has a much higher level of security compared to previously used mechanisms.

Introduction

The most common design mechanism of security mechanisms is employing complex mathematical problems whose computation outruns the rate at which the security mechanisms grow. The growth of Artificial Intelligence has become a challenge in which bots are programmed to detect subtle weaknesses or apparently unnoticeable patterns in the user’s security log-ins. In general passwords have remained the norm for securing the confidentiality of various avenues. The major challenge though remains the fact that passwords are vulnerable to online guessing attacks,, brute force attacks and reverse engineering mechanisms thus leading to a possibility of failure of security. Such a human machine interface (HMI) has culminated in the need for a security system which would need human intervention in accessing systems which would be competent enough to thwart back engineering or exhaustive searching algorithms thus making it a hard AI problem.

Conclusion

It can be seen from the previous discussions that the proposed technique achieves much higher level of randomness compared to previous techniques. The effect is the plummeting probability of breaking the algorithm by brute force or guessing attacks. It is evident that such a mechanism needs human intervention and hence pre programmed bots cannot be used to break the algorithm. Moreover employing SHA in place of MD5 enhances the security of the algorithm. Using AES/Blowfish for encryption of the data leads to much lesser vulnerability in eavesdropping or modifications in the coordinates of the captcha images. Thus the proposed technique can be used as an effective tool while dealing with hard Artificial Intelligence problems.