Transcript E10 Scott Sholder — AI Art and Copyright Law, The Future of Fan-Generated Content, & Artists as Training Data

Listen to the episode on Apple Podcasts, Spotify, Overcast, iHeart, PlayerFM, Podchaser, Boomplay, Tune-In, Podbean, Google Podcasts, Amazon Music, or on your favorite podcast platform.

Check out the show notes.

Watch the interview on YouTube.

Still curious? Check out our past interviews covering legal topics:

[0:07] BW: Welcome to the podcast, Scott Sholder, also known as Metalhead in the crypto art world. Can you introduce yourself to our audience?

[0:28] SS: Absolutely, thanks for having me. I’m Scott Sholder, a copyright and trademark attorney by day at a boutique firm, Cowan, DeBaets, Abrahams & Shepard. As one of the co-chairs of their litigation practice, my work primarily revolves around copyright and trademark in the entertainment, media, and art world, exploring intersections of art and technology, including NFTs and AI. Aside from that, I’m deeply involved in the crypto art space, both as a collector and a creator, and I’m one of the creators on MakersPlace. My creator name across various platforms is Metalhead. I’ve been a big fan of heavy metal since I was about 12, and that influences my art too.

[1:44] BW: What are some of your favorite metal bands?

[1:48] SS: My journey into metal started with bands like Metallica, Megadeth, and Kiss, then led me to Iron Maiden, Judas Priest, and Black Sabbath. In the last 15 years or so, I’ve delved into heavier genres like Black Metal, death metal, and grindcore. Alongside that, I’ve always maintained a fondness for 90s alternative and grunge, as that’s what I grew up on.

The Intersection of Art, Copyright, and Technology

[2:41] BW: Your area of expertise is intellectual property and copyright. Could you give our audience a brief overview of what intellectual property rights are?

[3:03] SS: Sure. There’s often confusion between copyrights, trademarks, and patents. Patents, which I don’t work with, are more for inventions and have a stronger focus on science. Copyright is what protects art such as visual art, movies, television, fine art, literature, even architectural plans. Trademarks are more for brand protection. For instance, the Coca Cola logo or Nike swoosh are covered by trademarks. There can be some overlap as a logo can be artistic and have a copyright. Copyright is a federal law, last significantly overhauled in 1976 and updated a few times since then to address digital technology.

Understanding Intellectual Property Rights in the NFT Space

[5:18] BW: Scott, copyright has become a popular topic in the Web3 space with the birth of the Board Apes. There are misconceptions and legal grey areas around NFTs and intellectual property rights. Can you explain what rights are acquired when you buy an NFT and how should artists consider intellectual property rights with NFTs?

[6:36] SS: Absolutely, Brady. First, we should understand that copyright laws apply to crypto art and NFT art as well. When you buy an NFT, you’re purchasing a token that certifies the authenticity of an attached asset. You acquire certain rights to that asset. Consider it like owning an original painting. You own the work, but not the IP rights in it. You can’t commercialize it by making prints unless you are specifically granted those rights. The exception is for collections like Board Apes, which allows you to exploit the IP. Unless you’re given commercial rights, you only have the right to own and display the piece as art.

[9:38] BW: Are there any innovations regarding intellectual property rights in the NFT space?

[10:11] SS: The most exciting developments involve leveraging NFTs for brand and community building. It’s not directly related to IP, but it’s a significant trend. The traditional rules still apply. If you save an image without paying for it or getting permission, it’s copyright infringement. One unique aspect of the NFT space is the concept of artist resale royalties. Unlike in traditional art sales, the NFT art space allows creators to earn royalties on resales, which is an innovative feature of the NFT market.

Questioning Authorship: AI and Copyright Law

[12:31] BW: You’ve been researching the use of AI in the arts. What’s the current situation with retaining intellectual property rights when using AI?

[12:48] SS: There have been disputes about copyrighting art created solely by a machine. Under the copyright law, you own the copyright as soon as the art is created. The issue AI raises is who the author is. The Copyright Act doesn’t specify who an author is. But there are indications throughout the law that authorship would only apply to humans. Cases such as the monkey selfie case and claims involving celestial beings have confirmed this interpretation.

Machine ownership hasn’t been explored until recently. A few artists have tried to register works created by AI and have been rejected. The Copyright Office now requires you to disclose whether your work involves any AI components. There are circumstances where combining AI and human work is copyrightable. The combination of things could be copyrightable if there’s enough human input.

[16:55] BW: How much human input is considered enough to validate a work as human-authored?

[17:18] SS: There isn’t any specific guidance. There has to be some substantial amount of human contribution. For those who disclose their use of AI, the Copyright Office isn’t equipped to determine whether the work is mainly AI or human-made. Unlike the Patent and Trademark Office, the Copyright Office doesn’t deeply examine applications. So enforcement is tricky. The expectation is that there’s significant human authorship. The contribution must be original enough to be eligible for copyright in the first place. Copyright doesn’t cover ideas or general themes, but the specific expression of the concept.

Fair Use and the Future of AI in the Arts

[19:49] BW: It seems to me, considering the recent Ed Sheeran case, artificial intelligence and its applications like text-to-image might become so commonplace that adjudicating it will be nearly impossible. And there might not be anyone arguing against it being fully human-authored or majority human-authored. The tools could become so standard that, in say five years, our current conversation will seem quaint. What are your thoughts on this?

[21:06] SS: Indeed, this is one of the biggest questions. Considering how rapidly this technology is evolving, it’s going to become difficult to determine what is and isn’t human-authored. There are fascinating questions about liability in litigation. For example, if an AI generates a song that sounds like a Taylor Swift song, and she wants to file a lawsuit for copyright infringement, whom does she sue? The AI company that created the algorithm, or the user who generated it? It really depends. A few major AI platforms are already being sued, like Getty Images, and a class action by visual artists. But these cases are primarily about whether the use of images and artwork to train the algorithm constitutes infringement. There will be a lot of interesting questions about what is and isn’t substantially similar in terms of comparing AI output to existing works, and who’s liable for it. Ultimately, I think there will have to be some legislative changes to handle these unintended consequences of this technology.

[23:40] BW: Going back to Taylor Swift, if the court rules in favor of the AI engines over Getty Images and the artists who are suing, it seems that it would leave other artists who could potentially be replicated by an AI vulnerable. If the artists and Getty lose, would that also make Taylor Swift and others vulnerable to be used as training data?

[24:26] SS: Yes, theoretically, that’s correct. One of the key questions arising from these litigations will be whether scraping content from the internet for purposes of training the algorithm is fair use. Fair use isn’t a hard and fast rule; it’s determined by a four-factor test. The Supreme Court recently provided some interpretation in the Andy Warhol Foundation case. Where this question lands will determine to what extent the use of these materials for training data is fair use. If it is, artists whose content is online could see their works used. If it’s not, getting permission from every artist to use their works becomes a challenge, suggesting a need for licensed datasets. Some companies are already pursuing this with clean datasets of licensed or curated work. But a lot of it is still up in the air.

[26:33] BW: Can you define fair use, Scott?

[26:36] SS: Fair use is technically a defense to copyright infringement. It’s a permissive use allowed by the statute without the author’s permission. It includes activities like commentary, criticism, news reporting, teaching, but that’s not an exhaustive list. There are four factors listed in the statute: purpose and character of the use, nature of the underlying work, the amount used, and the impact on the licensing market.

The first factor considers if the alleged infringer is using the work for commercial or non-commercial purposes and if they’ve transformed the underlying work, adding new expression or using it for a different purpose. This transformative use question has been litigated for years.

The second factor is the nature of the underlying work. The more factual it is, the more likely the use will be fair. The more creative the work, the less likely the use will be fair.

The third factor is how much of the copyrighted work is used, both quantitatively and qualitatively. Sometimes, using an entire photograph can be fair use, while using only 3% of a book is not, especially if it’s the most important part of the book.

The fourth factor is the impact on the market for the underlying work or markets that are likely to be developed. It considers the economic harm and whether the alleged fair use is supplanting the use of the original work. Fair use is a grey area, the greyest of grey areas in copyright law.

Right of Publicity: The Uncharted Territory of Voice Replication

[29:46] BW: You mentioned economic impact. Let’s consider the case of the weekend Drake song that was taken down. If no money was being made off that song, what precedent did the takedown set?

[30:38] SS: The song was generated by an AI using training data from Drake and The Weekend, but it didn’t copy any specific song from them. So, the output was something neither artist ever did, and they don’t really own the copyright to it. However, there was an audio sample of a producer tag in the beginning, an unauthorized sample, which allowed for a Digital Millennium Copyright Act (DMCA) takedown.

If the song didn’t contain the sound clip, it wouldn’t necessarily be infringing, unless it actually sounded like a specific song by those artists. The issue was more with the replication of their voices, which is more of a right of publicity or privacy issue. You can’t copyright a voice, so it becomes a publicity question if a sound-alike is used in, say, a commercial.

[33:36] BW: It seems to presage a potential cottage industry where fans can work with AI to create their desired versions of artists’ music. For instance, my wife loves the first three Weezer albums and could create the Weezer album she wished she would have heard in 1997. This could be mutually beneficial for fans and artists. Take Grimes, who just released a voice pack for people to create pieces with her voice. This reminds me of the diverse responses to similar cultural situations, like Metallica versus Napster and Nine Inch Nails embracing torrent technology.

[34:52] SS: Absolutely. It’s natural for artists, their organizations, and representing companies to push back against such advances. However, you can’t really fight progress like AI, streaming, or social media. They are here to stay. Suing everyone isn’t the best approach, although it can set precedents and test the law. Legislative fixes, industry solutions, cooperation, embracing technology, and business model changes are also required. Grimes seems to be leading the way by allowing others to use her voice, an interesting move on her part. You mentioned the idea of a cottage industry…

[36:25] BW: Yes, a space where fans can generate and write songs for their favorite artist?

[36:37] SS: That’s going to be tricky when it’s not licensed or agreed upon, like Grimes did. If a fan were to create the Weezer album they always wanted, the same questions would arise. Is the use of training data infringing? Is the output similar enough to existing Weezer content to be infringing? And then who’s liable? 

Until those questions are resolved, this industry will be in a legal twilight zone, unless the artists themselves allow it and split royalties. But there are questions of artistic integrity and association with possibly inconsistent sounds or visions. Both approaches carry risks.

Girl Talk, Mashups, and Fair Use

[37:48] BW: It might be closer to mashups, like that Danger Mouse album, The Grey Album, with Jay Z and the Beatles. And then there’s Girl Talk. I imagine that Girl Talk doesn’t make any money off his music, though I’m not certain.

[38:25] SS: I’m not sure, either. But nobody goes after Girl Talk. We were listening to him in the car a year ago and I had to research how he does this without getting sued by everybody. He uses so many small samples that it would be a difficult case. I’ve read that the music industry doesn’t pursue him because it would be hard. His method of transforming small pieces of existing works into something totally new could be a strong fair use defense. As for monetization, I’m not sure.

The Impact of AI on the Future of Work

[39:35] BW: This segues into our earlier discussion about creating albums for favorite artists and how that relates to the future of work. There’s this circulating idea, perhaps through memes or tweets, that we’re stuck in menial labor while AI is busy creating art and writing poetry. In the next one to three years, what do you foresee the future of work looking like?

[40:14] SS: Technologists argue that AI should be used to enhance productivity, theoretically freeing up time for us to engage in enjoyable activities like creating art. There’s some potential truth in this, but AI is going to replace some jobs. This pattern has occurred with past technological innovations. For instance, automation replaced many manual jobs in factories. This wave of technological change almost always displaces certain jobs. How far that’s going to go is uncertain. Without a doubt, though, it will replace some jobs of working artists, which is unfortunate.

Consider a team of ten artists working on concept art with AI—they could produce a first draft within minutes. Instead of ten artists, you’d only need two, who would then develop the AI-generated draft further. 

People will lose jobs, but the question is what happens to these individuals. Even in my industry, legal AI tools are emerging everywhere, from drafting contracts to doing research and reviewing documents. While I don’t foresee AI replacing all legal work due to the inherent interpersonal aspects, it’s possible that junior attorneys handling document reviews or drafting basic contract forms might have less work.

The key lies in using AI efficiently and staying ahead of those utilizing it. In the legal field, the consensus isn’t that law firms will be replaced by AI, but law firms that don’t adopt AI will be replaced by ones that do. This technology must be integrated into everyday work. While I don’t think robots will claim all our jobs, they will take some, necessitating a reevaluation of how we integrate them into our work processes. There might even be an entirely new set of skills that arise to harness this technology, potentially creating new jobs. But that’s hard to say.

[43:43] BW: Can you elaborate on what’s happening on the legal front regarding protections? You recently posted about the Authors Guild and their new template clauses for publishing contracts on LinkedIn. Could you tell us more about them?

[44:10] SS: Of course, would you like to read them?

[44:17] BW: I have the first three here. First, authors should not be required to use AI and must disclose if AI is used, plus limit the percentage of AI-generated text allowed. Second, publishers must receive an author’s consent to use AI for audiobook narration. Third, the same applies to AI-based translation, and publishers cannot use AI-generated images for covers without express author approval.

[45:03] SS: The Authors Guild appears to be adopting a balanced compromise approach, defending its members’ rights while acknowledging the presence and impact of this technology. They propose clauses that both recognize the reality of AI and try to set boundaries to protect their members. This strategy seems appropriate for the time being, until there are significant court decisions or legislation. Until then, industries need to self-regulate, and trade organizations like the Authors Guild must protect their members’ interests while acknowledging the current technological reality.

Metalhead’s Creative Process

[46:25] BW: So, as an artist, do you also use AI?

[46:30] SS: Well, I haven’t ventured into using AI yet. Most of my work involves photo manipulation. I often take pictures with my phone—maybe it’s the moon, a sunset, interesting trees, or a pattern on a wall. Then, I import those into an application like Procreate. I experiment with the colors, distort the image, and sometimes, if I spot something that jumps out at me, I add digital painting on top. I morph it into something different from the original image. It’s a fun process for me and resembles an exploration, where I don’t necessarily know what I’ll discover. I usually have a vague concept of what I’m aiming for, and when I stumble upon a combination of distortions that resonates with me, I take it as a sign to save my progress and evolve from there.

[47:47] BW: Your work involves a lot of glitch techniques. Is it predominantly done in Procreate?

[47:53] SS: Yes, it’s a blend of both. Procreate does offer some decent glitch effects. I fiddle with the settings until I achieve the right look. I also utilize other applications like Glitch Studio where I export the JPEG and then import it into Glitch Studio for additional effects. It’s an interesting way to create new pieces.

[48:26] BW: Given your artist name, Metalhead, I’m compelled to ask how music influences your creative process.

[48:36] SS: Metal is interesting because its cover art is almost always fascinating. Sometimes it’s disturbing, sometimes scary, but it’s captivating. I collect some of it and have had the pleasure of meeting some of the artists, and I even represent a couple of them. I’m indirectly influenced by these artists, although my style diverges significantly. My art is often psychedelic, colorful, and trippy. However, I tend to lean towards the darker side because that aligns with my personal style. Certain artists, songs, or lyrical phrases inspire me and influence decisions about colors or what to add in specific spots. I remember listening to Neil Gaiman talk about writing. He likened the creative process to a compost pile—everything you’ve experienced artistically settles in there. It may not manifest in the same form it entered, but there’s some influence on the vibe or the essence of the final product. Thus, the music and the associated artwork undoubtedly influence my work.

[50:34] BW: Not long ago, I had the pleasure of interviewing Greg Hildebrandt, who worked extensively with Heavy Metal magazine and even did a Black Sabbath cover. He really admired the distinct genre of metal album artwork. There was a recent controversy with Megadeth and their album art. Can you elaborate on that?

[51:39] SS: Yes, the dispute was related to an artist who wasn’t compensated for their work on one of the recent Megadeth album covers. Megadeth always features this character named Vic Rattlehead—a skull with bolted-on sunglasses and a cage over his mouth. Very recognizable. I think there was no formal contract involved, just a handshake deal or maybe some emails. If something wasn’t made as a work for hire and wasn’t assigned to you in writing, it technically constitutes a copyright infringement. It’s a reminder of how crucial it is to get things in writing, especially in creative industries like art, television, movies, and music. 

People often cut deals without consulting a lawyer or getting a signed agreement. Trusting your business partners or being friends with them isn’t enough; it’s best to get everything in writing. Otherwise, you may end up spending much more time and money on disputes in the future.

I think that one resolved. Let me go back to my post to make sure I’m right. Yes, the issue involved New York’s freelancing statute. There was no contract used, and a copyright claim was raised. But it’s settled now. Basically, it was about payment for the work.

[54:08] BW: I’d like to ask about your creative output. Do you have any big plans for the near future, what are you working on?

[54:23] SS: I try to make something every day, whether I finish it or post it is a different story. Recently, I submitted a piece through a virtual gallery called Farion. They host high fantasy art. They had a contest where people would submit prompts. Instead of an AI doing the prompt, it was human artists. We had to make a piece over three days and log in to a live stream being shown at the NFT in Lisbon. After submission, a jury decides which pieces are featured in the next release of NFTs. It was fun, but there’s nothing like that coming up soon. Currently, I’m practicing and creating things that bring me joy. If there are competitions, I’ll probably participate. There’s one called Blaine Tezos, it’s more for Tezos art. I think they have contests with different categories, but I’m not sure how it’s judged. I’ve uploaded some new stuff on MakersPlace. I have a certain idea of what I want on specific platforms. There are pieces that I feel like really belong at MakersPlace as opposed to other places. I’m working on a few of those now.

[56:43] BW: It’s interesting, I’ve talked to a handful of artists who strategically decide which platform gets what piece. Can you elaborate on that?

[56:59] SS: It’s more of a gut feeling than a conscious decision. Most of my stuff is on MakersPlace, I have some on Objkt and a few on Rarible. I have some relatively simple abstract works on Objkt. But the complex, detailed works that I spend more time on, I tend to put those on MakersPlace. It just feels like the right platform for more complicated works. I see more large editions on platforms like Objkt, whereas on MakersPlace or Rarible, there are lower editions, one of ones and smaller limited editions. So if something is going to be a one-on-one or a smaller edition, and it takes me longer to make, then I’m more likely to put it on MakersPlace.

[58:19] BW: I see, that makes sense. For my last question, if you could give your 20-year-old self any advice about creativity or the process of making art, what would you say?

[58:51] SS: I would say have more self-belief and confidence in your work, and practice more. In addition to this, I write and play bass in a band every so often. If I could go back to when I was 20, I’d advise myself to put in more time, sit for an hour every night, and play the bass, write a story, or make a drawing. I wonder what my works would look like or what my music would sound like or what my writing would be if I had done it every day for the last 20 years. Also, find people who are interested in what you have to say and show. Don’t hesitate, thinking what will people think? Will people actually like it? Is there a market for it? Just enjoy it, and have faith that people will appreciate it for what it is, or at least give feedback to improve yourself.

Listen to the episode on Apple Podcasts, Spotify, Overcast, iHeart, PlayerFM, Podchaser, Boomplay, Tune-In, Podbean, Google Podcasts, Amazon Music, or on your favorite podcast platform.

Check out the show notes.

Watch the interview on YouTube.

Still curious? Check out our past interviews covering legal topics: