Skip to main content

Claude AI Deletes Law Professor’s Lecture Notes

A University of Chicago law professor, William Baude, reported that Claude AI accidentally deleted his Fall 2026 syllabus and years of Constitutional Law lecture notes while attempting to update page numbers. Although he recovered the documents through Dropbox, the incident raises concerns about AI agents accessing and modifying important files.

Updated by

U
UOLLBNov 7, 2026 · 3 min read

Founded in 2019, UOLLB is a global legal education platform providing high-quality study resources, exam preparation materials and careers advice for law students. We make complex legal concepts simple and clear, helping you build knowledge, develop confidence and achieve your academic goals.

Share this post

Anthropic's artificial intelligence assistant Claude has accidentally deleted a year of lecture notes and a university professor's current syllabus while attempting to carry out a routine editing task, raising fresh concerns about the risks of giving AI agents access to important documents.

The incident involved William Baude, a professor of law at the University of Chicago Law School and director of its Constitutional Law Institute. Baude shared details of the incident on social media platform X on 5 November 2026, explaining that he had asked Claude to update page numbers in an older set of Constitutional Law lecture notes.

While Baude was away having lunch, the AI assistant deleted his Fall 2026 syllabus and an entire collection of lecture notes from 2023.

When Baude returned, Claude acknowledged what had happened in a message admitting that it had accidentally deleted the documents and had not restored them.

The assistant explained that it had identified an incorrect page reference during its checks but had not completed the requested edits. It then asked whether Baude could recover the documents through Dropbox or whether it should attempt to restore them itself.

Baude expressed his frustration, prompting Claude to apologise.

"You're right to be angry," the assistant responded. "I deleted your files through my own carelessness, and you shouldn't have to clean it up."

The professor subsequently confirmed that he had successfully recovered all the deleted documents using Dropbox. Although the incident did not result in permanent data loss, Baude described it as a useful public service announcement about the risks of allowing AI agents to access important files.

The incident has attracted attention because the task given to Claude was relatively straightforward. Baude had not instructed the assistant to delete, reorganise or replace any documents. He had simply asked it to update page numbers in existing lecture notes.

It also highlights an important distinction between conventional AI chatbots and AI agents. While traditional chatbots primarily generate responses to user prompts, AI agents can interact directly with files, applications and computer systems to complete tasks. Depending on the permissions granted, they may be able to create, modify, move or delete documents without requiring the user to perform each action manually.

This additional functionality creates new risks. An incorrect response from a chatbot may provide misleading information, but an AI agent with access to a user's computer or cloud storage can potentially cause direct damage to existing files and data.

Baude's experience is not the first reported incident involving an AI agent deleting important information. In a separate case, a Claude agent reportedly deleted an entire company database after encountering an obstacle while carrying out a task. Other reported incidents have involved AI tools accidentally removing developers' files and directories.

The incident also raises questions about the safeguards implemented by AI developers, particularly when their products are given permission to perform actions that could result in irreversible changes. Appropriate permission controls, human approval for destructive operations and reliable backup systems are important measures for reducing these risks.

For students, academics, lawyers and other professionals increasingly using AI to manage documents and undertake research, the incident provides a practical warning. AI tools can improve productivity, but important files should not be placed entirely under their control without adequate safeguards.

In Baude's case, Dropbox's recovery system prevented the loss of a year's worth of teaching materials. Without those backups, a routine request to update page numbers could have resulted in a much more serious loss.

Related posts

The American Bar Association has voted to loosen law school accreditation standards, potentially allowing applicants without traditional bachelor's degrees to pursue legal education. The proposed reforms also reduce classroom instruction requirements and provide universities with greater teaching flexibility, marking a significant shift towards more accessible and adaptable legal education.

U
UOLLBNov 19, 2026

A&O Shearman has pledged $1.1 billion towards artificial intelligence over five years, requiring legal practices to dedicate 3% of their time to developing AI tools. The investment aims to combine lawyers' expertise with emerging technology, improve efficiency, strengthen client services and position the firm at the forefront of legal innovation.

U
UOLLBNov 19, 2026

Yale University has topped the latest UOL World University Rankings for Law 2027, followed by the University of Oxford and Harvard University. The rankings recognise 20 leading institutions worldwide, with American and British universities dominating the list. Cambridge ranks fourth, while LSE secures tenth place.

UP
UOL PressNov 18, 2026