Policy4d ago
Policy
Claude 4.6 Grades Ancient Greek, Raises Integrity Concerns
Claude Opus 4.6 graded a Greek textbook’s problem set, prompting a user to worry about the model’s sycophancy. The user switched from grading to having the model write answers directly. This shift highlights concerns over AI‑assisted academic cheating. Educators must set clear guidelines to prevent misuse in academic settings today.