The increasing adoption of Artificial Intelligence (AI) in critical decision-making processes has highlighted the need for ensuring the security and reliability of AI systems. To this end, it is proposed the development of the Safe AI Framework, an evaluation system that assesses the overall health of an AI pipeline, including its Data, Model, and Environment components. Technically speaking, the evaluation system aims to identify potential vulnerabilities and provide suggestions for mitigation strategies to improve the health of the pipeline. The mitigation measures provided by the system can be categorized into data-related, model-related, and infrastructure-related mitigation measures. The proposed evaluation system is expected to provide several benefits to organizations, including the assessment of the security and reliability of their AI solutions, the identification of potential vulnerabilities, and compliance with regulations and standards related to AI.