| Yhteenveto: | This thesis looks at how Artificial Intelligence (AI), especially large language models (LLMs) like GPT and RoBERTa, can help improve cybersecurity by detecting threats and managing risks. Examples of these cyber threats include phishing and malware. Traditional systems that rely on fixed rules often don’t work well enough.
The study examines how well AI-based models can identify these threats in test settings, focusing on how quickly and reliably they detect issues. Data for training and testing the models came from sources like the PhishTank Phishing Database and the Common Vulnerabilities and Exposures (CVE) list. The results show that LLMs can accurately identify known threats, but they struggle with threats that are vague or uncommon.
This research gives insights into using AI in Information Security but notes that these technologies still need improvement to reduce false alarms and allow more customization for specific industries.
Keywords: Artificial Intelligence, AI, InfoSec, Information Security, Cybersecurity, Threat Detection, Large Language Models, Information Security, GPT, RoBERTa
|
|---|