AI Code Rewrite Challenges LGPL License

Post LinkedIn

🐯Read original on 虎嗅

#ai-coding #license-controversy #copyright-lawchardetchardet claude-code lgpl

💡AI code 'rewrites' threaten open-source licenses—key legal insights for devs.

⚡ 30-Second TL;DR

What Changed

Chardet v7.0 rewritten via Claude Code AI in days

Why It Matters

Could undermine GPL/LGPL ecosystems if AI rewrites legally evade copyleft, harming open-source reciprocity.

What To Do Next

Review LGPL compliance before using AI code gen on licensed libraries like Chardet.

Who should care:Developers & AI Engineers

Key Points

•Chardet v7.0 rewritten via Claude Code AI in days
•License changed from restrictive LGPL to permissive MIT
•Legal fight over AI bypass of open-source obligations

🧠 Deep Insight

Background and context from public sources — not the original article. 9 sources cited.

🔑 Enhanced Key Takeaways

•Maintainer Dan Blanchard created a detailed rewrite plan using Claude's 'superpowers brainstorming skill', starting with tests in an empty repository while instructing the AI to avoid basing code on LGPL sources[3][7].
•The new chardet v7.0 codebase shows only 1.29% structural similarity to the original, supporting claims of independence despite maintainer Dan's decade-long familiarity with the project[3][7].
•Chardet originated as a manual Python port by Mark Pilgrim from Mozilla’s MPL-licensed C++ universalchardet library around 2014 discussions on relicensing[7].

🔮 Future ImplicationsAI analysis grounded in cited sources

AI-assisted rewrites will be rejected as valid relicensing without verifiable clean-room isolation

Critics argue Claude's training on original code and maintainer influence violate clean-room principles, setting precedent against acceptance[2][4][7].

Copyleft licenses like LGPL will erode if AI relicensing succeeds

Accepting this method allows easy conversion of restrictive licenses to permissive ones by prompting LLMs to rewrite, undermining copyleft enforcement[2].