As a couple of others in the Alien Art Hive community, i use AI generative tools to create graphic artwork. Thinking of tagging and selling some collection under NFT, i wondered what was the legal framework surrounding my creations and began to dig the net.
Here are my notes.
NFT's
The minting and sale of NFTs are susceptible to copyfraud and infringement of copyright in the underlying work as well as the infringement of the moral rights of the author of the original work. Copyfraud can occur when a person mints an NFT of a public domain work, falsely claiming to own copyright in the underlying asset as an original work. Similarly, infringement of copyright and moral rights can occur when a person who is not the author or copyright owner in the underlying asset in which copyright subsists, mints an NFT and misrepresents that they are the author or copyright owner of the work. Such conduct is especially problematic in relation to NFTs due to the anonymity features of the blockchain, which make it difficult to verify who is the rightful creator or owner of copyright in the underlying work of an NFT.Source
Simply said, to be on the safe legal side of selling NFTs, i should make sure to be the author or the copyright owner of my work. It appears that in the case of generative art, authorship is not obvious. Who is the author ? Me or the programmer, or the AI algorithm, or the dataset owners ?
Well, in U.S. this question is debated the notion of "fair use" seems to open some doors to the artist using computer aided tools.
In applying this standard, courts consider four factors: (1) the purpose and character of the use, including whether the use is for nonprofit educational purposes; (2) the nature of the copyrighted work; (3) the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and (4) the effect of the use upon the potential market for or value of the copyrighted work. While the courts generally consider these factors on a case-by-case basis, "the number of factors that tip the balance in favor of fair use" will vary. "As a rule of thumb, courts are more likely to find fair use when the use is transformative, meaning that it changes the original work into a new form, with a further purpose or different character, which, by in large, does not substitute for the original use of the work"
But this is still a hot debate and the question so far remains open, so consequently the question of the copyright status should be taken in account:
What is the scope of copyright subsisting in AI-assisted work?
The scope of copyright in AI-assisted works might be fragmented along different lines in different jurisdictions. Traditionally, copyright laws have been based on human creators who control the creative process leading to an original work. But with the advent of AI systems, the role of such human authors seems to fade. In this respect, scholars who have discussed the protection of artificial creations made, for the most part, the assumption that any human creative input has become absent because of the (full) autonomy of the algorithms. As a result, they conclude that the works thus generated fall out of the copyright system. Yet, a deep look at the creative process reveals a collaboration between human artists and algorithms whose autonomy is in fact relative. Such atypical collaboration blurs the lines in copyright and makes it difficult to identify the author(s) of the resulting creation.Source
Moreover, i learnt that there is no worldwide consensus, countries by countries, the law might be different.
Countries differences.
U.K. legislation stipulates that when it comes to copyrights for AI generated work the protections will fall to the individual or group of individuals who created the software or made the needed arrangements for the AI generated work to be crafted. Depending on how much contribution an artist or programmer puts forth into creating the work will determine how much right they will have over it. Source
In other countries like Spain and Germany, it is not a matter of question because copyright protections are only offered to human authors exclusively.Source
In US, there are two ways in which a court can approach it. The first is that the copyright protections are not offered at all because the work is created by technology. The work would instead be deemed as public domain. The second option may be that a court can rule that the protections under the individual are at the heart of the chain of creation. If an artist and programmer got together to create an algorithm that generates a piece of work, then both those individuals would have joint copyright protections over the work. Source
India, Ireland, New Zealand, have taken the approach that authorship belongs to the programmer.Source
Additionally the question of the datasets rights used to build the AI models has to be taken into consideration, knowing that in our case, those datasets are made of thousands of pictures collected on the Net.
Datasets copyrights.
If a deep learning algorithm is trained on millions of copyrighted images, would the resulting image be copyrighted?
Right now, there is no copyright or licensing arrangement for using an author’s original writing as part of a machine learning model. We lack a consensus on how to judge the originality of a work essentially composed of random snippets of thousands, or even millions, of input works. Creative Commons advocate a cautious approach to copyright legislation, arguing that AI needs to be properly understood before any copyright implications can be addressed.Source
Without being a lawyer expert, let's have a quick look at licensing notice of some datasets currently feeding the popular models of a VQGAN+CLIP algorithm.
Imagenet : https://www.image-net.org/ : "The data is available for free to researchers for non-commercial use."
OpenImages : Attribution license," we make no representations or warranties regarding the license status of each image and you should verify the license for each image yourself."
Flickr : the following CC licenses allow commercial use: Attribution, Attribution-ShareAlike, Attribution-NoDerivs.
Note : Creative Commons licenses are a form of permission, but they have clearly defined limits. Stil all photos have copyright protection that lasts for over 70 years and consequently Commercial use is still requiring to give credit to the original photograph. But this is practically not possible.
Wikiart :WikiArt presents both public domain and copyright protected artworks.
To summarize, we can use, create and publish all of our artworks without legal risk, until there is no commercial usage.
Strictly legally speaking, it seems that a lot of dataset are mixing different kind of copyrighted and public domain source where a lage part of the data keeps their copyrights that can lead to infringment claims in case of commercial use. Altough a bit theoretical, the risk exist, especially for generative art which is quite new.
So yes, currently, in a conservative view, i am not sure we might consider any copyrighted ownership of our graphical artistics experiences when it deals with generative AI processes. But that may change in the future.
There is no question that AI software-created art holds value, yet the question as to how best to legally protect its creations has no real answer.Source
As illustration for this long quite tedious article, i used VQGAN+CLIP with an init picture related to this landmark in copyright suit involving the french photographer Patrick Cariou against Richard Prince.
Congratulations @dbddv01! You have completed the following achievement on the Hive blockchain and have been rewarded with new badge(s) :
Your next target is to reach 2500 upvotes.
You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word
STOPCheck out the last post from @hivebuzz:
This is something I'm constantly thinking about. Great post!
Thx @kaliyuga. forewarned is forearmed...
A question : Didn't you train your own model ? If so, i may imagine that if you can craft a dataset with pure public domain picture and collect all credits in a file that might be used as reference for commercial use... A bit tricky but it might be a temporary solution until better clarity is made in the future.
I've built a number of AI models, but right now I'm mostly working with a lot of things others built; Basically, I'm just not tokenizing any art I've made using models I didn't personally train. My instinct is that posting art made with others' models on Hive and being rewarded by the community isn't really the same as "commercial use," though, so I'm doing plenty of that with things I've made using VQGAN + Clip and Clip-guided Diffusion colab notebooks. Basically, I'm making a good-faith effort not to screw anyone over, and hopefully there will be clearer legal guidance soon :)
I agree, posting is not commercial use, there is no real trade, it's the post that is rewarded. This is not claiming copyright ownership...
Oh Clip-guided Diffusion ... did you succeed to get some results under a free Colab ? If yes i would be interested to know which notebook you found, because basically under a free Colab, i have to wait hours and hours to get a 256*256 size picture (usually i get a K80 with 12Gb GPU...). I wonder if there are some optimized notebook that would require less resources to play with Guided diffusion, because the produced pictures looked quite more polished and detailed than VQGAN. So far i did not find.
I've been using the 256x256 HD Rivershavewings notebook, and it does take a remarkably long time, but it's a more resource-intensive process than VQGAN. I'm pretty sure a less resource-intensive notebook using the same underlying architecture would produce much less accurate results. I find the diffusion notebook more than worth the wait, honestly; I've gotten some incredible results using it. 256x256 is definitely small, but I upscale my outputs using an ESRGAN notebook by @jotakrevs. Successful prompt engineering for diffusion is different than for VQGAN+Clip, too. Just by feel alone, I'd say VQGAN feels more like sculpting with words/poetry and Diffusion feels more like building/architecture. I don't know if that makes sense.
Thx for your answer. I think we are indeed aligned. If i find alternative useful resources, i'll share it.
Please do!!!
You both had an interesting talk here! I cannot access the notebook due permissions... maybe you shared the link that only works with people you add to the list.
Diffusion, first time I hear about. I'm eager to test this stuff 😛
Here's a link to the tweet announcing the notebook! That should work for you :)
I think you'll really love Clip-Guided Diffusion. It produces more realistic/accurate results than VQGAN+Clip, and diffusion models outperform GANs in general, as well :)
Hi @jotakrevs. First off, sorry to butt in like this in this comment thread. I have seen amazing pictures of clip guided diffusion by RiversHaveWings on her twitter feed. Just look at those images and you'll be tempted to try it out. Thank you @kaliyuga for posting the links to the notebooks. I too will check it out later.
This is for those who have Colab Pro, needs 16 GB of video memory :
One day the AI's will be fully independent by humans..and it will be in their interest to enforce copyright...
Great post...it's the first time I read about this topic!
!discovery 30
!PIZZA
@dbddv01! I sent you a slice of $PIZZA on behalf of @lallo.
Learn more about $PIZZA Token at hive.pizza (1/10)
law is made to be broken ;-)
that's for sure :)
This post was shared and voted inside the discord by the curators team of discovery-it
Join our community! hive-193212
Discovery-it is also a Witness, vote for us here
Delegate to us for passive income. Check our 80% fee-back Program
Thank you !
I kept having similar doubts regarding AI art and thought that it is better not to mint till there are clearer guidelines. Thanks for the elaborate post.
Congratulations @dbddv01! You received a personal badge!
You can view your badges on your board and compare yourself to others in the Ranking
Check out the last post from @hivebuzz: