In this assignment, I trained an image classifier in Teachable Machine to tell the difference between a person wearing a visible employee badge and a person without one. I connected AI to cybersecurity by using the model as a possible secondary check against tailgating, where an unauthorized person follows an employee through a secure door and gets around physical access controls. My first model kept picking up on “something near the neck” instead of the badge itself, so I revised it with more training images, including white objects like socks and string, until it focused on the badge’s shape rather than shallow cues like color and position. This showed me how much the quality of training data shapes what a model actually learns.