The thesis underscores the effectiveness of artificial intelligence in combating phishing attacks and underscores the importance of continued research and innovation to safeguard users and their data in the digital realm. Beyond technical aspects, the paper provides a theoretical framework on cyber security, phishing attacks, and the role of artificial intelligence in modern cyber defense. It emphasizes the ongoing need for research and improvement in this area due to the evolving nature of cyber threats. This BSc theis work was done by Mr Ivan Novakovic under the menthorship of prof. Armin Alibasic.
ABSTRACT – This thesis addresses the critical issue of detecting phishing URLs, a prevalent cyber threat, using advanced artificial intelligence techniques. It examines three models—Logistic Regression, Random Forest, and LightGBM—through rigorous analysis and real data testing to assess their effectiveness in identifying malicious URLs. The results demonstrate that these models, when properly trained and calibrated, can achieve high accuracy in predicting the maliciousness of URLs.