A new tutorial demonstrates how to build an AI detector to serve as a verifier for training small language models. The project explores how SLMs can learn to bypass detection filters. It serves as an educational exercise in the limitations of current detection tools. Practitioners can use this to test model robustness.