Generative AI and Security Operations Center Productivity: Evidence from Live Operations

TL;DR

This study investigates how adopting generative AI tools impacts security operations center productivity, finding a significant reduction in incident resolution time using real-world operational data.

Contribution

It provides empirical evidence linking GAI adoption to improved security incident resolution efficiency using observational data from live operations.

Findings

30.13% reduction in mean time to resolution

Robust results across various models

First observational study on GAI's impact in live security operations

Abstract

We measure the association between generative AI (GAI) tool adoption and security operations center productivity. We find that GAI adoption is associated with a 30.13% reduction in security incident mean time to resolution. This result is robust to several modeling decisions. While unobserved confounders inhibit causal identification, this result is among the first to use observational data from live operations to investigate the relationship between GAI adoption and security worker productivity.