Open Source Maintainer Uses AI to Rewrite Library and Change License, Sparking Debate on Future of Software Licensing
Key Takeaways
- ▸Chardet maintainer Dan Blanchard used Anthropic's Claude AI to rewrite the library and changed its license from LGPL to MIT, claiming less than 1.3% code similarity with previous versions
- ▸Original creator Mark Pilgrim disputes the license change, arguing that exposure to the original code and LGPL requirements make the new license invalid regardless of AI involvement
- ▸Bruce Perens warns this case demonstrates how AI could fundamentally undermine software licensing frameworks, particularly copyleft licenses like GPL/LGPL
Summary
Dan Blanchard, maintainer of the popular Python library chardet, has released version 7.0 under an MIT license instead of the previous GNU Lesser General Public License (LGPL), claiming he used Anthropic's Claude AI to create a clean room implementation that shares less than 1.3% similarity with prior versions. The move has ignited controversy in the open source community, with original creator Mark Pilgrim arguing that the license change violates LGPL requirements regardless of whether AI was used for the rewrite. Blanchard contends the AI-assisted rewrite is legally distinct from the original codebase and was done to improve performance by 48x and enable potential inclusion in Python's standard library.
Open source advocate Bruce Perens has characterized this dispute as evidence that AI will fundamentally undermine software licensing frameworks. The controversy raises critical questions about whether AI-generated code constitutes a derivative work when the AI may have been trained on the original licensed code, and whether maintainers can use AI tools to effectively circumvent copyleft license obligations. The chardet library receives approximately 130 million downloads per month, making the licensing decision highly consequential for the Python ecosystem.
The dispute highlights an emerging legal gray area where AI code generation intersects with software licensing. While Blanchard used plagiarism detection tools to demonstrate structural differences between versions, questions remain about whether Claude ingested chardet's code during training and whether that would constitute copying under copyright law. The case may set precedent for how the software industry handles AI-assisted rewrites of open source projects, potentially threatening the enforceability of copyleft licenses that have been foundational to the open source movement.
- The controversy raises unresolved legal questions about whether AI-generated code trained on licensed software constitutes a derivative work under copyright law
- With 130 million monthly downloads, the chardet licensing decision has significant implications for the Python ecosystem and could set precedent for AI-assisted open source development
Editorial Opinion
This dispute represents a potential inflection point for open source licensing in the AI era. If maintainers can use AI to effectively 'launder' copyleft obligations through rewrites, it could fundamentally undermine the legal frameworks that have sustained community-driven software development for decades. The core question—whether AI training on licensed code and subsequent generation constitutes copying—urgently needs legal clarity, as the answer will shape the future relationship between AI tools and open source ecosystems. Courts and legislators must address this ambiguity before it creates a flood of similar licensing disputes.
