Our Select Committee continues to think & consider deeply about users' rights in the advent of AI Programming Assistance Software, such as machine learning models trained on large codebases. More in the recent committee minutes: lists.copyleft.org/pipermail/a


Does "Copilot" claim ownership of its contributed code to Github/Microsoft? Or is it regarded as "public domain"/"non-copyleft+free"?

ISTM that the process of AI generation might be regarded as diluting, in the same way that a person who "learns programming" does not have their output bound by the license of the code examples they learned from.

Yet unlike the student programmer, the AI is not a legal person to whom a new copyright would adhere.



"Perhaps that which cannot be easily attributed to a single source should be regarded as public property" is a sound principle for such things.

There is a similar debate around AI generated art (MidJourney, Craiyon...) currently, which I think raises the same issues of authorship and license.


It's also similar to the "ape selfie" case in which a non-human (ape) took his own picture with the photographer's camera. The court apparently agreed with the photographer's claim of ownership for providing the camera, though this would NOT be true, had the ape instead been another human. AIs, like apes, are not permitted to own things themselves, in our legal framework, creating a similar problem.

But did the court conclude correctly (legally and/or morally) in that case?

Sign in to participate in the conversation

Mastodon.ART — Your friendly creative home on the Fediverse! Interact with friends and discover new ones, all on a platform that is community-owned and ad-free. (翻译:DeepL)mastodon.art是艺术家和艺术爱好者的空间,而不是政治内容的空间--有许多其他的fediverse实例,你可以加入以获得更多的一般内容(而且你仍然可以从任何实例中关注你在.art上的朋友);见https://instances.social :)