Like a Hammer, It Can Build, It Can Break: Large Language Model Uses, Perceptions, and Adoption in Cybersecurity Operations on Reddit

The article presents a mixed-methods analysis of discussions from cybersecurity-focused Reddit forums regarding the use and perception of large language models (LLMs) in Security Operations Centers (SOCs). Analyzing 892 posts, the study identifies patterns in LLM adoption, revealing that practitioners utilize LLMs for low-risk, productivity-oriented tasks while expressing concerns about reliability, verification, and security risks that limit their autonomy. The findings underscore the need for careful development and integration of LLM tools to enhance efficiency without compromising organizational security.