Cornell University instructors are adopting analog assessment methods to address AI-written work, including oral exams and typewriter-based writing exercises. In one account, a biomedical engineering instructor requires an in-person “oral defense” without devices, framing the test as a response to students using AI tools. Separately, a German language instructor introduced a semester practice using manual typewriters with no delete keys, online dictionaries, or spellcheckers. The goal is not only to slow output but to force students to demonstrate thinking and language production directly. These classroom-level changes reflect a broader institutional response to generative AI: shifting evaluation from production artifacts to documented reasoning and real-time demonstration.