The open source community is grappling with a contentious issue after chardet version 7.0 was released using AI assistance. Developer Dan Blanchard used Claude Code to completely rewrite the popular Python library, claiming it's a fresh, MIT-licensed implementation that's 48x faster than previous versions. The original library, created by Mark Pilgrim in 2006 under an LGPL license, has strict redistribution requirements that Pilgrim argues should still apply.

Blanchard maintains that the AI-generated code is structurally independent of the original, citing similarity analysis showing only 1.29 percent overlap between versions. He created what he calls an "AI clean room" process, starting with an empty repository and explicitly instructing Claude not to base anything on the original LGPL-licensed code. However, Pilgrim contends that extensive exposure to the original codebase and Claude's training on similar code makes this a derivative work requiring the same LGPL license.

This dispute highlights broader questions about AI-generated code and open source licensing. With courts yet to rule on AI authorship for software, the legal status remains unclear. As one observer noted, "the nature of software changed" when AI made rewriting projects so simple. The outcome could have far-reaching implications for how open source projects are maintained, licensed, and potentially relicensed in an AI-assisted development future.