What happened: The maintainer of the Python library chardet released a new major version under the permissive MIT license, replacing the previous LGPL license, and argued the change is justified because the new code was effectively a clean-room rewrite produced with the help of Anthropic’s Claude.

Why it matters: If “rewrite with an LLM” becomes a practical path to shed copyleft obligations, it could undermine the economic and governance assumptions that have held open-source licensing together for decades — and force courts and communities to decide what counts as copying in an AI era.

Wider context: The dispute sits at the intersection of three unsettled areas: whether training is legally “copying,” whether AI-assisted code has enough human authorship to be copyrightable, and whether exposure to prior code defeats clean-room claims even if a tool generated the new implementation.

Background: The maintainer cited very low similarity scores between the new release and prior versions using plagiarism-detection tooling, while critics argue the project can’t escape LGPL obligations if the maintainer had prior access to the older codebase. The Register also quotes open-source veterans warning that cheap reimplementation changes the software business model.

Chardet dispute shows how AI will kill software licensing, argues Bruce Perens — The Register

Singularity Soup Take: “LLMs can rewrite it” isn’t just a legal headache — it’s a trust problem. If communities believe copyleft can be bypassed on demand, the long-term effect won’t be cleaner licensing; it’ll be fewer ambitious shared projects worth trying to protect in the first place.

Key Takeaways:

• License shift at issue: chardet moved from LGPL to MIT in version 7.0, raising questions about whether a maintainer can relicense a project when prior contributors and terms exist.
• Clean-room claim via AI: The maintainer argued the new release is a rewrite created with Claude’s help and claimed low structural similarity to earlier versions, though critics say prior exposure to the code undermines the clean-room premise.
• Copyleft pressure test: Open-source leaders quoted in the piece argue that if software can be cheaply reimplemented from behavior or tests, copyleft’s enforcement leverage weakens — potentially reshaping how both open and proprietary software are funded.

Related News

How creators and coders are fighting the AI slop tide — The same core tension: when generative tools reduce the cost of production, the fight shifts to provenance, attribution, and incentives.