A professor got caught using ChatGPT to write his own lectures. He would've never been caught if he used this Harvard professor's NotebookLM workflow. Here's what gave him away and the exact prompts that make it undetectable. He never opened ChatGPT and typed "write me a lecture on thermodynamics." His first prompt was always: "Here are my syllabus, my past 3 lectures, and my research papers. Find the concepts my students struggle with most in week 6. Then build an outline that connects those gaps to the assigned readings in my voice." He wasn't asking AI to write for him. He was asking AI to think like him. But the move that made me close my laptop and stare at the ceiling was his second one. He uploaded two years of student feedback forms into NotebookLM. Then asked: "Find the pattern in where students felt lost. What am I explaining poorly and how would I fix it based on my own writing style?" Every other professor was using ChatGPT as a ghostwriter. He was using NotebookLM as a mirror. His third prompt was saved as a shortcut on his phone. "Here are my past lectures and this week's material. Write 3 discussion questions that sound like they came from someone who has taught this course for a decade." Three prompts. Every single week. ...