Finetuning Large Language Models for Security Log Detections

Explore the innovative application of finetuning Large Language Models (LLMs) for security log detections in this 25-minute conference talk from BSidesSF 2024. Delve into the limitations of traditional security log detection methods that rely on pre-defined signatures, and discover how LLMs can be leveraged to create more sophisticated and generalizable detection systems. Learn the process of finetuning popular open-source LLMs for specific security log detection use cases, potentially revolutionizing the approach to identifying and mitigating security threats. Gain insights from speaker Wilson Tang on how this cutting-edge technique can enhance cybersecurity practices and improve the accuracy of threat detection in log data.