인공지능을 통한 보안 및 복원력 강화: GenAI가 예측 AI에 합류

This IDC Perspective discusses how IDC defines AI in cybersecurity. Combinations of artificial intelligence (AI) and machine learning (ML) have influenced the cybersecurity landscape for the better of 15 years. What computers have always been able to do is make correlations to the bytes, files, hashes, and code that comprise a network. However, for all of the improvements in computing and years of refining algorithms, so much of operating the network and the cybersecurity software that protects the network are still manually intensive processes.After all of this time, recent developments in generative AI and, more specifically, ChatGPT are seemingly addressing the cybersecurity manpower gap. IDC calls it "autonomizing the SOC." The process of realizing a fully autonomous SOC involves several intermediary steps, but the new efficiencies in evidence are:The availability of enriched data at the time of the incident investigationThe ability to generate an instantaneous response based on the type of attack is increasingly automatedThe implementation of analytics to discover unmanaged devices in the networkThe development of natural language processing (NLP) that enables threat hunting and security querying at the speed of speech"In creating security outcomes, the hard work is not about the artificial intelligence but creating and enabling the security data foundations that will allow AI to create measurable benefits to IT professionals. Regardless, the cause for overall optimism is real," says Frank Dickson, group vice president, IDC Security and Trust.

Executive Snapshot

Situation Overview

• Defining Artificial Intelligence
  • Definitions
  • Artificial Intelligence as Applied to Security
  • User Behavioral Analytics - Somewhere Between Machine Learning and Artificial Intelligence
• It's Not About the AI; It's About the Data
  • Data Framework Structures
  • Data Management
    • The Role of AI in Data Resilience
  • Data Curation
• Concerns and Considerations in Leveraging AI
  • Present Limits to AI
  • Hallucinations and the Role of the Analyst
  • Data Security and Privacy Risks
    • To Presort with Labels or Without Labels Creates a Question of Privacy
  • Input/Content Manipulation/Bias
  • Efficacy/Seed Set

Advice for the Technology Buyer

Learn More

• Related Research
• Synopsis