Chardet dispute shows how AI will kill software licensing, argues Bruce Perens

Steven

Chardet dispute shows how AI will kill software licensing, argues Bruce Perens

Summary

Maintainer Dan Blanchard has published version 7.0 of the Python library chardet under an MIT licence after using Anthropic’s Claude to produce a purported ‘clean-room’ rewrite. The project was previously under the GNU LGPL, and the change has provoked a dispute: an individual claiming to be the original author argues the LGPL prevents relicencing, while Blanchard says automated similarity checks (JPlag) show less than 1.3% overlap with prior releases.

Prominent developers and advocates are split. Some welcome the permissive licence and performance improvements (Blanchard reports a 48x speed boost and hopes for inclusion in the Python standard library). Others — notably the Free Software Foundation and Bruce Perens — warn that AI-trained models that can reimplement code threaten copyleft enforcement and the economics of software licensing more broadly.

Key Points

  • Blanchard used an LLM (Anthropic’s Claude) to help produce a rewrite of chardet and changed the project from LGPL to MIT.
  • An individual claiming to be original author Mark Pilgrim opened a GitHub issue contesting the licence change and arguing exposure to original code prevents a clean-room defence.
  • Blanchard cites JPlag results showing under 1.3% similarity to prior versions; critics question whether the model was trained on the original code and whether that matters legally.
  • The debate highlights unresolved questions about how much AI assistance dilutes human authorship and whether AI output can be treated as independently copyrightable.
  • Bruce Perens warns the episode could upend the economics of proprietary and copyleft software; the Free Software Foundation calls undermining copyleft ‘antisocial’.
  • Possible outcomes: more permissive relicensing, legal challenges, shifts in how software is monetised, and changes to policies on AI-assisted code in foundations and language ecosystems.

Context and relevance

This isn’t just a repo spat — it’s a concrete example of how capable code-focused LLMs are changing authorship and licence enforcement. The matter ties into broader legal and policy debates (for example recent US cases on AI-generated works) and affects developers, maintainers, companies relying on copyleft protections, and organisations drafting AI training rules or contributor policies.

For Python users the stakes are practical: chardet is widely used (~130M downloads per month) and a faster permissively licenced version could be adopted quickly, while the precedent could encourage relicensing of other projects or provoke coordinated legal responses.

Why should I read this?

Because this tiny library kerfuffle might change how you can use, share or charge for software. If you write, audit, ship or licence code, the rules you’re used to could be about to shift — and fast. Read this to get ahead of a likely wave of relicensing, policy changes and legal fights that will affect development strategy and compliance.

Source

Source: https://go.theregister.com/feed/www.theregister.com/2026/03/06/ai_kills_software_licensing/