AI in Cybersecurity

Abstract

The intent of this project is to address the importance of artificial intelligence in cybersecurity by describing its various applications in both attacks and defense. As artificial intelligence finds a place in more and more aspects of life, we must be aware of its potential strengths and shortcomings, especially from a security standpoint. Artificial intelligence suffers from one crucial dependency above all else - it is only as effective as the data used to train it allows. This detail alone allows for two distinct approaches to undermine its integrity; namely, the use of methods or information that lie outside the bounds of that training data, or the manipulation of the training data itself to create vulnerabilities in the AI. Despite these shortcomings, however, AI technology presents a means of detecting and preventing intrusions that far outstrip human ability alone. A great deal of currently available antivirus software is itself a rudimentary implementation of AI, analyzing patterns and signatures in software and external data to prevent intrusion. Machine learning, in particular, is an incredibly versatile method for analyzing valid or invalid data and making appropriate judgements regarding system security.