Prompt Guard

A lightweight prompt injection detection classification model released by Meta, trained specifically for jailbreak and injection attack scenarios. It can be integrated into AI applications as a security defense layer.