Simon WillisonMar 5
The chardet open source library relicensed from LGPL to MIT two days ago thanks to a Claude Code assisted "clean room" rewrite - but original author Mark Pilgrim is disputing that the way this was done justifies the change in license - my notes here: https://simonwillison.net/2026/Mar/5/chardet/
Can coding agents relicense open source through a “clean room” implementation of code?

Over the past few months it’s become clear that coding agents are extraordinarily good at building a weird version of a “clean room” implementation of code. The most famous version …

Simon Willison’s Weblog
Show threadǝʌɐpMar 5
@simon lots of public conversation logs for the courts to pour over when the time comes
Show threadǝʌɐp
@simon I think I'm of the opinion that it's impossible to clean room something that is open source with an LLM because the exact source is in the training corpus
Mar 5 at 5:00pmWeb
Show threadCassandrichMar 5

RE: https://social.treehouse.systems/@dvshkn/116177633529242686

@dvshkn @simon Bingo.

Show threadMagnus AhltorpMar 9

@dalias @dvshkn @simon Yes, and this phrase is especially clueless:

”explicitly instructed Claude not to base anything on LGPL/GPL-licensed code”

Is there any evidence, or even indication, that putting this into the ”instructions” has any effect whatsoever?

Show threadCassandrichMar 9
@ahltorp @dvshkn @simon Nope. 🙃
Show threadTimMar 5
@dvshkn @simon A clean room implementation isn't *required* for something to be non-infringing though, it just makes it easier to prove that you didn't infringe.
Show threadmirthMar 6
@dvshkn @simon One could imagine using a model so small it is provable that it is not capable of reproducing significant chunks of the input. Of course, there may be some echoes of the architecture or whatever, but that seems fair in general whether done by hand or with autocomplete. Whether such a model could be made is an open question.