With the rapid growth of mobile internet usage, phishing attacks targeting smartphones have become more sophisticated, exploiting users through malicious URLs embedded in emails, SMS, and social media. This thesis presents a deep learning-based model designed to detect and mitigate URL-based phishing attacks on smartphones, enhancing personal data security.The research explores the limitations of traditional phishing detection methods and highlights the need for AI-driven solutions capable of adapting to evolving threats. By leveraging neural networks and machine learning techniques, the proposed model analyzes URL structures, content features, and behavioral patterns to accurately classify phishing links in real time.Additionally, this study evaluates the model's performance against existing cybersecurity solutions, demonstrating its effectiveness in identifying phishing URLs with high accuracy and low false positive rates. The findings contribute to the development of advanced, AI-powered mobile security systems that protect users from cyber threats without compromising device performance.