Dan Blanchard, the developer of the chardet library for character encoding detection in Python, has released a new version of the library under a new license. Version 7.0 uses the MIT license instead of the LGPL license. The developer claims that the AI assistant Anthropic Claude, now listed as a contributor, rewrote the library without using any original code, allowing him to replace the copyleft license with a permissive one.
A user identifying himself as Marc Pilgrim, the original creator of the library, joined the discussion on this action (unfortunately, this user's account is empty, making it impossible to verify whether he is who he claims to be) and created an issue stating that Dan has no right to change the project's license.
Dan countered that he compared versions 7.0.0 and 6.0.0 using a plagiarism detection library, finding that the similarity between the versions was less than 1.3%. According to him, no file in the codebase of version 7.0.0 resembles any file from any previous release.
Dan explained that he wanted the chardet library to be included in the Python standard library, as it is a key dependency for many Python projects. However, three obstacles hindered integration: licensing, speed, and accuracy. As a result of the work done, the speed of encoding detection by the library has increased by 48 times. For a project that is widely used in many other projects, this optimization will significantly improve performance for millions of users (the package is downloaded about 130 million times a month).
Zoe Kuijman, Executive Director of the Free Software Foundation, told The Register, "We cannot comment on the specifics or legality of this project without further research or consulting with lawyers, but there is nothing 'clean' about a large language model (LLM) using code that it is required to rewrite."
In this regard, Zoe decided to give a moral assessment of Dan's actions: "As for the essence of the GPL, a permissive license is still technically a free software license, but undermining copyleft is a serious act. Refusing to grant others the rights that you yourself received as a user is extremely anti-social, regardless of the method used."
Bruce Perens, the author of the original definition of open source, expressed concern about the situation: "The upcoming social changes are equally significant and, as we are already seeing, even more alarming."
The legal implications of using a large language model to repurpose both open and proprietary projects (from which the source code has been stolen) are difficult to predict and highly dependent on precedents set by the courts. However, if a large language model rewrites code into a programming language different from the original, it becomes nearly impossible to prove the creation of a derivative work.
Recently, the U.S. Supreme Court declined to review the case 'Thaler v. Perlman', in which the plaintiff attempted to overturn a lower court's decision stating that he could not copyright an image created by artificial intelligence. The U.S. Supreme Court upheld the lower court's ruling that works entirely created by AI are not protected by copyright.
Source: opennet.ru
