In the Spring 2026 offering of CS 240: Programming in C, the instructor enforced a syllabus‑wide ban on using ChatGPT, Claude, Copilot or any LLM to generate assignment solutions. To detect violations, a static‑analysis tool named Argus was created; it searches for code indicators that are virtually absent before 2024 and rise sharply thereafter, producing an H‑score with high accuracy. Argus was applied to the semester’s submissions, flagging 267 of 584 students (≈45.7 %) as showing strong evidence of AI‑generated code. Each flagged student received an email requiring completion of a self‑report form by April 20 2026 5 pm EDT; the form asked them to acknowledge responsibility for the specific assignments and accept a zero on those tasks. Of the 207 enrolled students contacted, 117 responded, with 108 admitting guilt and nine contesting the findings. Subsequently, 144 students dropped the course, of whom roughly half (74) later re‑enrolled. Performance analysis showed that students identified as AI users scored 10.75 % lower on the first midterm and 14.5 % lower on the second midterm compared with peers. After concerns about coercion and due process, the instructor decided to disregard Argus’s retroactive findings, using the tool only for remaining assignments and future course offerings while allowing dropped students to re‑enroll.