Template talk:PD-algorithm

Jurisdiction

While the analysis given in this template is certainly true for the United States[1] and most other countries, it isn't true for the United Kingdom, which has a special limited term of copyright protection for AI-generated works of 50 years from creation, with the author being "the person by whom the arrangements necessary for the creation of the work are undertaken".[2] Nosferattus (talk) 19:09, 9 July 2022 (UTC)

Legal basis

@BMacZero: , I am writing to you as the creator of this banner. In it, it is written that "his file is in the public domain because, as the work of a computer algorithm or artificial intelligence, it has no human author in whom copyright is vested".

What is the legal basis for this statement? There is no reference to this in the banner, nor in the documentation. The only reference we can find is that this is not the case in the UK.

Thanks, Dirac (talk) 15:19, 19 January 2023 (UTC)

@Dirac: Nosferattus has linked a relevant ruling from the copyright office above. You can also see in the copyright office's "Copyright Basics" circular that "An original work of authorship is a work that is independently created by a human author"[3]. – BMacZero (🗩) 22:27, 19 January 2023 (UTC)

@BMacZero: , two things come to mind when reading your answer:

1. Shouldn't be explicitly written that it comes from these documents and that it applies to the US? It may make sense because the Foundation's servers are in the US, but it still needs to be explicit I think. I have no idea what the status is in Canada for example. Or do we assume that if it is ok in the US, we don't need any other information?
2. Don't we need a secondary source, which clearly indicates what it means for the images created by AI in the US, and which refer to these documents? Here I find myself interpreting a judgment, on the one hand, and deducing that if it's not made by a human it's not copyrighted, on the other hand. Who am I to make these inferences? Wouldn't it be better to have an authoritative reference that makes this interpretation, and not to have each contributor make the interpretation?

What do you think of this? Dirac (talk) 16:05, 20 January 2023 (UTC)

@Dirac: Well, for Commons, the copyrights that matter are the United States copyright and the copyright of the work's country of origin. However, I would be okay with making this a US-specific template and renaming it to something like {{PD-US-algorithm}}, since the theory is not authorized by some kind of international treaty, but subject to whatever the laws are in the country of origin. Having a big list of countries where this template is and isn't okay would be unwieldy as well, and that would avoid it.

As far as AI-generated images, yes, they are a big deal now so it's probably a good idea to add a line explicitly addressing them. The copyright office ruling posted by Nosferattus above deals with an AI-generated artwork and should be a relevant source. – BMacZero (🗩) 18:23, 21 January 2023 (UTC)

@BMacZero: thanks for your answer. I think a more explicit banner would effectively be welcomed given the current context about AI. From what I understand from your answer, you clearly have more expertise than I have to perform this. How could this be done? Dirac (talk) 23:49, 21 January 2023 (UTC)

I will try to find some time to do this. The template needs to be moved and language needs to be added to note that it is US-specific. – BMacZero (🗩) 17:25, 23 January 2023 (UTC)

@BMacZero: this decision from the US Copyright Office is more recent (February 24th 2023) and covers this aspect explicitly. Hope this helps, Dirac (talk) 14:40, 27 February 2023 (UTC)

@Dirac: Thanks, that's another good source. Coming back to this now, I'm not as confident making this change to this template that's now used fairly widely. I'm going to try to solicit more input at COM:VP. – BMacZero (🗩) 03:54, 1 March 2023 (UTC)

Repeating what I said at Commons:Deletion requests/File:Alice and Sparkle cover.jpg: "My !vote would be to just ignore source country altogether for AI images and follow US law only, like we do for COM:PD-Art. I think the reason for requiring an image to be free in its source country is that it intuitively "belongs" to that country and it's the most important country for it to be free in. So for old artworks, for example, we do care what country the work itself is from, but not where it was digitized. However, AI art is almost all published online to an international audience, sometimes by an anonymous Internet user from anywhere in the world, making it have much weaker ties to any particular country than most other images. Try identifying the source country of this: An artwork generated by an AI model, developed by an American company, on that company's European servers in Ireland, prompted by a Brazilian living in the UK uploaded to Wikimedia's Dutch servers while on holiday in France. (But like the case of PD-Art, source country still applies to COM:DW concerns, so for example an American cannot ask DALL-E 2 to produce an image of the Tour Montparnasse and upload it to Commons.)"

Adding to my statement: We've generally never had to worry about "source country" of Internet works before. Either it's a potentially PD-old work, in which case the Internet didn't exist back then, or it's a human-created work published on the Internet which is almost always a copyvio if not freely licensed. So we haven't had to worry about "source country" since the freeness of Internet works didn't depend on it. All that has changed with generative AI.

Therefore, I see no need to move this template. Like PD-Art, this will be a case where we apply only US law to all works regardless of origin. -- King of ♥ ♦ ♣ ♠ 00:50, 3 March 2023 (UTC)

I agree with King of Hearts. The source country requirement doesn't make any sense for AI-generated works and is completely impractical for them anyway. We should only be concerned with U.S. law in this case, which is relatively straightforward (ignoring the complication of derivative works for now). So I think we only need one licensing template for PD-algorithm, but it should include notices for reusers in countries that provide copyright protection for AI-generated works (so far, just the UK). This country-specific information is for re-users, however, not uploaders, and it is equally valid for works from any country. For example, if I create an AI-generated work in the United States (using a model hosted in China that publishes to a website hosted in Germany), it is theoretically protected by a 50-year copyright within the UK. Nosferattus (talk) 20:00, 4 March 2023 (UTC)

Murky waters here

I'm sure I don't need to tell you all just how much controversy there has been lately of AI image generators copying actual artists (due to overfitting etc.), and I think there is a genuine argument to make, or at least a notable concern, that some AI generated images might therefore be considered derivative works of copyrighted material and thus very much not public domain. I personally do not have a definitive take on this matter. Really, I'd just like to start up a discussion on this. OmegaFallon (talk) 22:00, 6 March 2023 (UTC)

@OmegaFallon: See Commons:AI-generated media#Can AI-generated media be lawfully hosted on Commons?. Nosferattus (talk) 02:34, 7 March 2023 (UTC)

Wording

The current wording of this template is:

"This file is in the public domain because, as the work of a computer algorithm or artificial intelligence, it has no human author in whom copyright is vested."

However, while in its Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence the US Copyright Office has stated that if

"a work’s traditional elements of authorship were produced by a machine, the work lacks human authorship and the Office will not register it"

it has also stated that

"In other cases, however, a work containing AI-generated material will also contain sufficient human authorship to support a copyright claim. For example, a human may select or arrange AI-generated material in a sufficiently creative way that “the resulting work as a whole constitutes an original work of authorship.” Or an artist may modify material originally generated by AI technology to such a degree that the modifications meet the standard for copyright protection. In these cases, copyright will only protect the human-authored aspects of the work, which are “independent of ” and do “not affect” the copyright status of the AI-generated material itself."

Imho the wording of the template should be clarified to make this clear. I would suggest something like:

"This file is in the public domain because it is the work of a computer algorithm or artificial intelligence and does not contain sufficient human authorship to support a copyright claim."

Jaqen (talk) 18:46, 17 July 2023 (UTC)

Completely agree, it's the correct way to tell it -- g (talk) 18:51, 17 July 2023 (UTC)

I agree as well. Gestumblindi (talk) 18:57, 17 July 2023 (UTC)

I don't see any problems with that. – BMacZero (🗩) 02:27, 18 July 2023 (UTC)

 Done The Squirrel Conspiracy (talk) 23:46, 28 July 2023 (UTC)

This should nowadays say "Public Domain in the United States"

As the recent China court ruled that AI-generated artworks can be protected by copyright in China. Liuxinyu970226 (talk) 01:14, 3 December 2023 (UTC)