Microsoft Azure blog inadvertently guides users to pirate Harry Potter for LLM training
A Microsoft developer blog post on using LangChain with Azure SQL's vector store included instructions that effectively guided readers to pirate Harry Potter books for use in LLM training examples. The post drew 170 points and 93 comments on Hacker News.
The gaffe highlights the casual approach some tech companies take toward copyrighted training data, even as the industry faces mounting legal challenges over AI training on protected content.
View full digest for February 19, 2026