Can coding agents relicense open source through a "clean room" implementation?

9 hours ago

Coding agents can perform 'clean room' implementations of code, a process historically done by teams of engineers over weeks or months, in just hours.
The chardet Python library, originally LGPL-licensed, was rewritten and relicensed as MIT by a maintainer, sparking debate over the legality and ethics of such actions.
The maintainer argues the rewrite is structurally independent of the original, supported by low similarity percentages from plagiarism detection tools, and was done with AI assistance without direct access to the original code.
The original creator disputes the relicensing, citing LGPL violations and questioning the independence of the rewrite given the maintainer's long-term exposure to the original code.
The case raises broader questions about AI's role in creating fresh implementations of existing code, with implications for both open source and commercial IP.
Potential future litigation is anticipated as companies become aware of the threats to their closely held IP from such AI-driven implementations.

Hasty Briefsbeta