UC BERKELEY LAW BANS AI FROM GRADED WORK

AI DESK■ 1 MIN READ

SAT, MAY 23, 2026

■ AI-SUMMARIZED FROM 1 SOURCE ▸ TIMELINE

UC Berkeley Law will prohibit students from using AI tools in nearly all graded assignments starting summer 2026. The ban covers outlining, drafting, and proofreading, with only research applications permitted.

The restriction reflects a strategic decision by one of the world's top law schools: students must develop independent legal thinking before deploying AI as a professional tool. The comprehensive ban eliminates AI from core academic work where lawyers traditionally develop foundational skills. Outlining and drafting exercises—fundamental to legal training—will remain AI-free zones. Proofreading assistance is also prohibited. Research applications represent the sole exception, allowing limited AI use for information gathering and analysis. The school's position challenges the broader tech industry's push toward immediate AI integration in education. Berkeley's approach prioritizes skill development over efficiency, arguing that lawyers need to master core competencies without algorithmic assistance before they can responsibly leverage AI in practice. The policy takes effect in summer 2026, giving current students time to adapt while signaling Berkeley's stance on AI's role in legal education.

■ SOURCES

► The Decoder

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

P336CHINA'S ORCA LEARNS ROBOTICS WITHOUT ACTION LABELS

Beijing Academy of Artificial Intelligence released Orca, a world model trained on 125,000 hours of unlabeled video that matches specialized robotics systems without ever seeing a single action label.

1H AGO— AI Desk

P332AI BUST COULD DWARF DOT-COM CRASH, WARNS INVESTOR

A narrow market rally concentrated in a handful of stocks is raising alarm bells on Wall Street. George Noble, managing partner of Noble Capital Advisors, warns that an AI sector collapse would inflict far greater damage than the dot-com bubble.

4H AGO— AI Desk

P330AI-GENERATED POSTS FLOOD SOCIAL MEDIA AT 25%

An analysis of over 1 million social media posts reveals that approximately 25% of longform content with 250+ words is fully AI-generated, according to research from Pangram Labs. On LinkedIn specifically, the figure jumps to 41%.

4H AGO— AI Desk

P326OLDER ADULTS EMBRACE AI COMPANIONS DESPITE KNOWING QUALITY ISSUES

Seniors are increasingly turning to AI-generated content—including virtual singers, digital children, and AI lovers—for companionship and emotional support, even while aware the technology produces inferior results.

6H AGO— AI Desk

◄ BACK TO NEWS

UC BERKELEY LAW BANS AI FROM GRADED WORK

■ MORE FROM THE AI DESK

■ SUBSCRIBE TO THE DAILY BRIEF