AI can rewrite open source code, but can it also rewrite the license?

There are computer engineers and programmers was long based on reverse engineering as a method of copying the functionality of a computer program without directly copying the copyrighted code of the program. Now, AI coding tools are raising new challenges about how that “clean room” rewriting process is done—legally, ethically, and practically.

With the release of the new version last week, these issues came to the fore charda popular open source python library for auto-detecting character encoding. The repository was originally written by coder Mark Pilgrim In 2006 and released under LGPL license imposed severe restrictions on its reuse and redistribution.

Dan Blanchard took over maintenance of the repository in 2012, but his release caused some controversy. chardet version 7.0 last week. Blanchard described the overhaul as a “major, MIT-licensed rewrite” of the entire library built using Claude Code to be faster and more accurate than before.

who spoke to The RegisterBlanchard said he’s long wanted to add chardet to the Python standard library, but hasn’t had time to address the “licensing, speed, and accuracy” issues that make it difficult to achieve that goal. With the help of Claude Code, Blanchard says he was able to overhaul the library “in about five days” and get a 48x performance boost to boot.

However, not everyone is satisfied with this result. A poster using the name Mark Pilgrim It appeared on Github to argue that this new version amounts to an illegal relicensing of Pilgrim’s original code under the more permissive MIT license (allowing, among other things, its use in closed-source projects). As a modification of the original LGPL-licensed code, Pilgrim claims that this new version of chardet should retain the same LGPL license.