Advancements of technology in the field of networks and computing bring with it a heightened importance of cyber security. Currently, legal forms of identification, financial information, scientific data, and state records rely on this technology - and as we become more dependent on cloud computing, data-base storage, and online banking, greater measures of keeping information safe is paramount. Because of this cyber-attacks will have the ability to inflict more damage on a larger demographics and will become more sophisticated and complex as technology advances. Because of this, a more robust method of securing information against poor programming, interpreted protocols, and design loop-holes will need to be developed in the event of cyber-attacks. Here we will demonstrate the feasibility of using a recurrent neural network as a method of tracking changing methods of cyber-attacks as they evolve with better camouflaging techniques. We introduce the concept of 4D as a method of securing online information: deter, detect, delay, and defend. Creation of a server called “The Honeypot” will be analyzed for cyber-attacks. The implementation of our defense mechanism will be performed using a recurrent neural network with deep learning model to make it a more robust protection against such attacks.