Coding agents cannot reliably "clean room" relicense open source — legal and technical gray zone¶
Insight: The chardet library relicensing dispute (LGPL to MIT via AI-generated "clean room" rewrite) exposes a legal gray zone: coding agents trained on open-source code cannot guarantee truly independent reimplementation. Even with JPlag similarity analysis showing only 1.29% code overlap, the maintainer's decade-long familiarity with the codebase and the AI model's likely training on the original code undermine clean-room defensibility.
Detail: Dan Blanchard (chardet maintainer since 2013) released v7.0.0 as MIT-licensed, claiming AI-generated clean room implementation using Claude without LGPL exposure. Original creator Mark Pilgrim contested the relicensing. Complicating factors: Blanchard's deep familiarity creates inherent bias; Claude was likely trained on chardet code; the original chardet was itself ported from Mozilla's MPL-licensed C library; reusing the same PyPI package name weakens the defense. Willison frames this as "a microcosm of the larger question around coding agents for fresh implementations of existing, mature code" and anticipates future commercial litigation.