Abstract: A system can receive a guardrail policy request that specifies a guardrail policy to assess for deployment on a server to protect at least a specific port of the server. The system can execute a fingerprint clustering machine learning model using server fingerprint data to generate cluster data that identifies a virtual machine cluster that includes a plurality of virtual machines executed by the server. The system can execute a traffic discovery machine learning model using server traffic data and the cluster data to generate a confidence score indicative of whether deployment of the guardrail policy would have an adverse impact on the server. The system can execute a risk assessment machine learning model using the application type data to generate a risk assessment score. The system can evaluate the confidence score and the risk assessment score and can determine whether the guardrail policy should be deployed on the server.

Abstract: Methods, systems, and apparatuses, may manage machine learned security for computer program products, which may create dynamic micro-perimeters.