Learning to Secure AI

Author: Adam Kohnke, CISA, CISSP, PNPT
Date Published: 22 August 2019

The trends appear to be presenting themselves all over the place; TV commercials, online ads, corporate product announcements, etc., are all saying the same thing: Artificial intelligence (AI) adoption and use are exploding. As an information security and assurance professional, I admit that I did not really know much about this emerging technology, so I decided to begin the process of becoming educated on the subject, even if only at an introductory level. I started performing online research to understand the current market size, future growth projections, how to achieve certification and education and, most important, approaches to governing and securing use of AI solutions.

My company presently allocates each employee a modest annual training budget, so I leveraged those funds to select a training provider and begin taking AI classes as I performed my research for my recent Journal article. I gravitated towards edX as their curriculum was 100% free, but also provided certificates after completing courses and quizzes, which is also useful for IT certification continuing professional education (CPE). As I completed my AI edX courses and online research, I wanted to structure my ISACA Journal article in a conversational and informative matter starting with defining AI and addressing some common misconceptions. From there, I wanted to address market size, projected growth trends and who the players are in the market. I believe this is always important because this information provides important context on what to expect in the near term and long term and which organizations to keep your eye on.

This biggest challenge I encountered was when it came to research and learning how to secure AI solutions. While AI is not new (it has been researched, discussed and developed over the past several decades), it has only become commercially adopted within the past 10 years and is still in its infancy. I did not find a free-to-use, mainstream security framework or set of publications that discussed how to directly approach AI security. Through several online articles and the completion of the introductory edX courses, I managed to Frankenstein the article together, and my hope is that at least 1 reader will learn something valuable that will assist or empower their enterprise in securing the use of AI solutions. My secondary challenge is, if you can write a related article or audit program, please do. It will benefit us all!

Read Adam Kohnke's recent Journal article:

"Preparing for the AI Revolution," ISACA Journal, volume 4, 2019.