As cybersecurity threats grow, the telecom industry faces significant challenges in ensuring web security, particularly against vulnerabilities exploitable by malicious actors. Traditional rule-based security systems struggle to keep pace with evolving threats, as their manual updates often delay the detection of new attack types. This work focuses on developing a machine learning (ML)-based model to address telecom vulnerabilities affecting web security. Using supervised learning techniques and labeled historical data, the model can identify known attack patterns and predict future vulnerabilities, ensuring defense against well-understood threats. Additionally, unsupervised learning enables the model to detect new, unseen attack patterns within the data, enhancing adaptability. Unlike classical systems requiring frequent manual updates, the ML model evolves automatically, improving anomaly detection with minimal human intervention. This dynamic, automated approach is scalable, efficient, and applicable to various telecom environments, outperforming traditional rule-based methods in detecting novel threats. The research demonstrates the effectiveness of ML-driven solutions in identifying unnoticed vulnerabilities and enhancing telecom infrastructure security. While this study develops and tests the model, future work will focus on optimizing algorithms, incorporating more datasets, and testing under varied conditions to improve performance and assess real-world applicability. This lays a foundation for advancing ML-based methods in telecom cybersecurity, particularly in detecting new vulnerabilities.