📄 Abstract
As cyber threats are becoming increasingly complex and scaled, it is difficult for traditional security mechanisms to keep pace in terms of timeliness, context awareness, and adaptiveness. The emergence of LLMs, particularly transformer-based architectures, has expanded the horizons for cybersecurity, thereby enabling appropriate threat detection, real-time analysis, and response automation. This study discusses several practical use cases of AI-driven security with the help of LLMs in three major verticals: financial services, healthcare, and critical infrastructure. Each illustrates unique but successful approaches toward implementing LLM-enabled threat analysis. In the financial sector, a cloud-based LLM anomaly detection system demonstrated a 45% reduction in false positives and improved incident response time by 60%. In the case of healthcare, the integration of LLMs with SIEM systems demonstrated a 38% avoidance of undetected phishing attempts and dwell time reduced by 30%. Critical infrastructure operators demonstrate proactive defense against zero-day vulnerabilities with the potential to achieve mitigation rates up to 70% faster through the use of LLM-powered threat intelligence. Through a comparison of these implementations, this research paper underlines how LLMs, when combined with other cybersecurity frameworks, can turn security operations into proactive, adaptive, and efficient processes. Key findings support AI-driven threat detection as a significant enabler of resilient digital ecosystems in the light of pervasive cyber risk.
🏷️ Keywords
📚 How to Cite:
Prabgun Mokha , Dr.Archana Kumar , AN AI-DRIVEN APPROACH TO CYBERSECURITY: USING LLMS FOR THREAT DETECTION AND ANALYSIS , Volume 11 , Issue 11, November 2025, EPRA International Journal of Multidisciplinary Research (IJMR) , DOI: https://doi.org/10.36713/epra24892