Today's highly skilled attackers exploit many of the vulnerabilities in their network. On the other hand, the risk of data leakage has increased dramatically because software or application vulnerability is always left without a fix. Such vulnerability using the (Zero Day), hackers will receive the target network, and can steal sensitive data. Detection of zero day traditional defenses is difficult, because the signature information zero-day attacks are unknown. Therefore, the need for new security solutions, that detect zero-day attacks, and evaluate the identified zero day vulnerability severity. The paper proposed an approach for detecting unknown vulnerabilities. The system is a framework that is a comprehensive approach for detecting and prioritizing zero-day attacks and removes these detected attacks. The proposed framework is based on probabilistic approach, to determine the Zero-Day attack path, and the subsequent degree of severity of the identified zero-day vulnerability. It is a hybrid of detection and removal method based on the detection of unknown defects present in the network, which is not yet detected. The system also shows the file with original size and with an attacked size.