Content area
Full Text
Abstract: Increased growth in the use of AI is lacking a cybersecurity and privacy framework In the paper AI secure development is introduced along with AI Dev/Sec/Ops which leads into the creation and understanding of having an AI cybersecurity framework for ML, DNN and CC systems. AI deviations are examined along with twenty AI cyber security issues which require the cybersecurity community to become learned and develop mitigations. The AI Cybersecurity framework addresses threat forecasting and risk trees as well AI system hardening and continuous monitoring. The paper can be used by cybersecurity professionals to start implementing an AI cybersecurity program to ensure AI systems meet the security and privacy requirements of the system throughout the AI-DLC of the system.
Keywords:artificial intelligence, AI cybersecurity framework, AI development life cycle, AI dev sec ops, AI deviations, AI cybersecurity issues
1. Introduction
Artificial intelligence (AI) is currently being implemented into information systems across all industries. AI is viable in the professional and personal realm of our lives and will provide benefits and efficiency. In the report, "The Future of Employment: How susceptible are jobs to Computerization," researchers from Oxford, Frey and Osborne, stated US workers would lose jobs to automation by 47% in the next two decades (Rouhiainen, 2018). Additionally, the Fourth Industrial Revolution will utilize a plethora of new technological advances and AI revolutionizing technology across multiple industries (Schwab & Davis, 2018).
The purpose of this paper is to introduce an AI Cybersecurity Framework which can be used to secure AI systems. To best understand the implementation of the AI Cybersecurity Framework, the researchers developed and presented the AI development life cycle (AI-DLC) and the AI Dev/Sec/Ops model. Both models are a part of AI development involving securing the AI system, the code, and ensuring the algorithm is viable. The paper also covers AI systems which went rogue, behaved inappropriately from the intended design. The paper will also cover 20 identified AI cybersecurity issues. AI is integrated into information technology environments creating a current need to ensure the AI systems meet cybersecurity and privacy requirements for being secure and hardened. Current and new AI technology will require the cybersecurity professional to rethink the standard cybersecurity approach and require a deeper understanding of AI and how...
