In recent years, artificial intelligence (AI) has made significant strides, primarily because of the widespread adoption and usage of open source machine learning models across various industries. Given the high resource demands of training the models with large datasets, many applications now rely on pre-trained models which save considerable time and resources, allowing organizations to concentrate on training and sharing these crucial models. However, using open-source models introduces risks issues in privacy and security that are often neglected. These models can sometimes harbor hidden functionalities that, when specific input patterns triggered , can alter system behavior, such as causing self-driving cars to disregard other vehicles. The impact of successful privacy and security attacks can range from minor disruptions in service to extremely serious consequences, which lead to outcomes such as physical harm or the disclosure of sensitive user information and data. This research offers an in-depth review of common privacy and security risks linked to open-source models, aiming to raise awareness and encourage the responsible and secure use of AI systems.