Summary
The maintainer of Python's chardet library, Dan Blanchard, released version 7.0 as an AI-generated rewrite using Claude LLM, switching from LGPL copyleft to permissive MIT license while keeping the same package name and API.[3][1][2] Original author Mark Pilgrim objected, arguing the process wasn't a true clean-room implementation and violates LGPL terms.[3][2] The move, aimed at improving speed, accuracy, and Python standard library inclusion, has sparked debate on AI's role in undermining open source licensing enforcement.[1][4]
Key Takeaways
- Chardet 7.0 is an AI-rewritten version of the LGPL library, now MIT-licensed with the same PyPI name and API for drop-in replacement.[1][2]
- Maintainer Dan Blanchard used Claude LLM to achieve in days what would take teams weeks, targeting Python stdlib inclusion.[3][4]
- Original creator Mark Pilgrim contests the relicensing, citing lack of clean-room conditions and LGPL persistence.[3][2]
- Debate centers on whether AI-generated code from licensed inputs creates independent authorship or inherits restrictions.[1][4]
- Impacts include calls for updated OSS licenses to address AI training and output constraints.[1][6]
Balanced Perspective
Chardet 7.0 was created by feeding the original LGPL code and tests into Claude AI, producing a structurally independent rewrite with identical API under MIT.[1][4][2] Blanchard acknowledges it's not a traditional clean-room process due to his prior knowledge, but claims measurability of independence supports relicensing.[4] Legal validity remains untested, with no lawsuits filed yet, though it highlights unresolved questions on AI outputs' copyright status.[1][3]
Optimistic View
This breakthrough demonstrates AI's power to accelerate maintenance of critical libraries, delivering a faster, more accurate chardet in just days instead of months, benefiting millions of Python developers.[3][4] Permissive MIT licensing removes barriers to commercial adoption and standard library integration, fostering broader innovation and collaboration.[3] Ultimately, it proves AI can democratize clean-room reimplementations, making high-quality software accessible without outdated licensing hurdles.[4]
Critical View
AI 'rewrites' like this erode copyleft protections, allowing maintainers to bypass LGPL obligations without genuine clean-room separation, potentially contaminating downstream projects.[3][1] Original author Mark Pilgrim's objection underscores risks of infringement, as exposure to source code taints the process regardless of AI involvement.[3][2] This sets a dangerous precedent, threatening the open source ecosystem's foundational licensing models and inviting litigation chaos.[3][5]
Source
Originally reported by theregister.com