Key Points

• Berkeley Law moved from limited AI permission to a stricter exam and coursework policy that bars generative AI in exams and sharply limits it in credited work.
• Edinburgh's philosophy department adopted a default no-generative-AI rule for assessed work, while allowing only specified proofreading and citation tools that meet university standards.
• Both policies emphasize academic integrity, foundational writing and analysis skills, and the need to preserve assessment validity as AI tools improve.
• Enforcement is a recurring constraint: schools can police submissions and exams, but cannot fully prevent AI summaries or AI-assisted outputs inside research platforms and search tools.
• The policy trend is not uniform across all teaching contexts; instructors and AI-focused courses retain some ability to authorize alternative rules with notice or disclosure.
• The signal is concentrated in higher education governance rather than broad government regulation, suggesting a practical, local response to generative AI's impact on assessment.