AI-Generated Code Can Evade Copyleft Licensing, chardet Dispute Reveals

Available in: 中文
2026-03-29T20:28:50.042Z·1 min read
The maintainer of Python's `chardet` library has used Anthropic's Claude to create a clean-room rewrite — with under 1.3% structural similarity to the original — and relicensed it from LGPL to MIT....

The maintainer of Python's chardet library has used Anthropic's Claude to create a clean-room rewrite — with under 1.3% structural similarity to the original — and relicensed it from LGPL to MIT. The dispute exposes how AI-assisted development could undermine copyleft licensing.

The Dispute

Why It Matters

If AI can produce functionally equivalent code with negligible structural similarity to the original, copyleft licenses become unenforceable. The LGPL requires derivative works to remain under the same terms — but if an AI generates new code from a natural language description of the same functionality, is it a derivative work?

The Practical Outcome

Bruce Perens' Take

Open source pioneer Bruce Perens argues this shows how AI will kill software licensing. The fundamental question: if AI ingested copyleft code during training, does its output infringe? If the output is structurally distinct, current legal frameworks have no answer.

Armin Ronacher's Observation

Flask creator Armin Ronacher noted: "Software licensing and the laws around it haven't been tested a lot in this new world of AI-assisted development."

Source: The Register

↗ Original source · 2026-03-29T00:00:00.000Z
← Previous: Anthropic Economists Rework AI Job Impact Measurement: 'Observed Exposure' Shows Minimal Labor Disruption So FarNext: Meta and International Law Enforcement Disrupt Major Southeast Asian Scam Networks, Arrest 21 Suspects →
Comments0