The BrXnd Marketing X AI Conference is coming to NYC on 5/16. Only a few tickets left. Get yours now! →

Copyright versus Generative Creativity

The Intellectual Property (IP) Questions Around AI in Marketing

A discussion on the changing world of intellectual property and what marketers and creatives need to know.

Transcript

Hello, my name is Colin, and this is Shannon, and we realize that this is kind of a heavy subject matter, copyrights, the law, but I wanted to start off with something fun and a little bit more pop culture, so will you explain to us both what happened with Drake, The Weeknd, and artificial intelligence at a high level, and then also let's get into some of the details about the legal implications, and we can use that as a jumping off point.

Definitely. Well, thank you to Noah, and thank you, Colin. It's rare that people are excited to talk about copyrights, so I'm happy to be here. What happened with Drake and The Weeknd, so on April 4th, this person who was identified as a ghostwriter, and they've said, I'm a ghostwriter for major labels, and I haven't gotten enough credit for my work. I've been writing for these labels, and I haven't been given any of that sort of fame or whatever that all of the big celebrities are getting, so that's their motivation. They write this song. They put the vocals through an AI generator to really, really accurately mimic the voices of Drake and The Weeknd, and the song quickly goes viral on TikTok, on YouTube, and they put it on Spotify and streaming services, so it was a little bit different from the kind of things that you're seeing where people are mimicking Kanye or having Kanye sing Hey There Delilah or Smash Mouth, and you see those on TikTok, and they're funny, and those might kind of fall into this parody camp of maybe that's fair use, but the difference here is they put it on streaming services, so they would get paid for that, and the problem is you're appropriating someone else's voice to make money, so the major labels are very nervous about this, right? Universal represents both The Weeknd and Drake. They get it taken down pretty quickly once it starts to get attention, and the interesting thing is they were not able to take it down because of copyright. They were able to take it down, or sorry, they were not able to take it down because of the right of publicity, so when you're mimicking someone's voice or their sound or their likeness, that's a publicity right, and it's not the same. It kind of falls in the same camp as intellectual property rights, but it's different than copyright, but they were able to get the work taken down because of this copyright technicality where the lyrics had included some already copyrighted lyrics, so it opens up some pretty big questions about what are major labels and music publishers going to be able to do if people are making money or making songs appropriating Beyonce or Taylor Swift, and if they don't have that copyright hook that they had in this case, what are they going to do? So there's a lot of big questions, and I think they fall into sort of the publicity and the copyright buckets, and those kind of all go together. And we were talking over there about Grimes had to walk something back. Explain what happened there. So Grimes, after this all happened, gets on Twitter and goes, I think it would be really cool to be fused with the machine, so please feel free to use my voice, and then in the next tweet she goes, actually, feel free to use my voice, but give me 50%, and then in the next tweet she goes, sorry, sorry, now my lawyers are weighing in, feel free to use my voice, but not for anything wildly offensive, you know. So it's this interesting question of if you're using someone else's publicity, right, even with their consent, how far are you going to let that go? And what's funny about that is it was clear that she had kind of like an interesting provocation of like giving her thoughts on technology in the future and stuff, but then it was like, and the more I've thought about this, the more I've opened up a can of worms, right? So yeah, because it's a cool concept, and then there's a lot of big questions. And that's what's important with this conversation, is we have a room full of people in the creative worlds, in the brand world, marketing world, builders, and we want to experiment with new things, but we also need to know when we can go off-piste and when we can't go off-piste. So next thing I wanted to talk about is, so there's two parts of AI IP. There's like the input side, which is the corpus of things that you train something on, right? And then there's the output side of like something goes out into the world and the copyright. I would love for you to kind of just talk about those two modalities. Yeah, so when you're looking at sort of the copyright concerns on the infringement side, it has a lot to do with this exclusive right to reproduction and distribution. So if I own a copyright in a song, I have the exclusive right to reproduce it, I have the exclusive right to distribute it, and I have the exclusive right to make or authorize other people to make derivatives, like remixes and arrangements, things like that. So what's happening on the input side is you have data aggregators, web crawlers or scrapers, that are sending bots out and they're scraping everything on the internet, and they're putting it into these big data repositories, and then AI is trained using that sort of corpus of data. And to the extent that it includes, and we know that it often does include copyrighted material, there's a really big question as to whether that violates copyright owners' rights to, you know, reproduce and distribute their works, but also have that derivative component of if you're using it to make something else, are you getting permission and consent for that? So that's on the input side, and it flows really naturally into the output side, which is if an output, if an AI generates something that includes a copyrighted, anything that's copyrighted, like an AI generates a song and you're including a snippet of lyrics or, you know, a sequence of notes that was found somewhere in the training data, you might end up with an infringing derivative work on the output side. And then the other really big question that's not so specific to input or output, but like in the copyright world what we're talking about, is this bad for human creators, you know, and how do we harness these tools without creating a system where human creators put work out into the world and then AIs take it, learn it, put out new material that competes with humans at a cheaper and faster pace. So the instant devaluation of human creativity. Yeah. So we want to figure out how to preserve human artists. What's interesting is so much of music culture in the past has been predicated off of like samples, right? Taking a break beat, speeding it up, making a drum and bass track, right? So we've had so much interesting precedent with snippets of intellectual property creating something new, but there's a system by which those originators are compensated. Yeah. Yeah. If you're sampling something, they, you license it. You know, Ariana Grande famously released Seven Rings, which has like, borrows really heavily from Rodgers and Hammerstein, Sound of Music, and the Rodgers and Hammerstein guys got 90% of the publishing for that song because of how heavily it was sampled. The funny story about that with Trent Reznor ended up making money off of Old Town Road because there was a loop from a Nails record, right? Yeah. Yeah. Yeah. Noah alluded to, but you're probably one of the only people that's read all of the new, you know, US Copyright Office guidelines. I have read it. Give us the TLDR on that for those that aren't as legally minded. Yeah. So the Copyright Office put out this guidance. It was in response to a case where they had to reject a copyright application because someone said that they were the author, but really there was an AI author. It was a comic book case. The text was written by a human. The photos were written, were generated by an AI, and the Copyright Office said, we'll you the copyright in the text and in the selection and arrangement of the photos, but we can't give you a copyright in the photos or in the, not photos, in the images because they're not human authored. So the guidance came out right on the heels of that. In a nutshell, it is copyright only protects human creativity and human authorship. So if it's something that's purely AI generated, it can't receive copyright protection. And is there nuance in this? Because as we're, a lot of the conversations today, there is an interesting poetry that's happening between humans and AI with prompts and nudging many touch points, a little dance back and forth. So in those instances where it's not a one click output to AI, but there's a little bit more back and forth, is there nuance there to unpack? I think so. And I think this is going to be one of those big questions that we will probably see answered over the next several years is like people take cases to court or challenge the Copyright Office's determinations. But the way that the guidance is now is that if you're not interacting with the material basically after it comes out of the AI, that's AI generated, even if you have a lot of interaction and input on the front side, similar to how if you're commissioning a work from a painter and you say, I want it to have generally this color scheme with a lot of reds and greens and I want a mom and a baby and a tree, but the artist is the one ultimately executing that and there's a lot of different ways that could be executed. In the same way, even if you put a lot of direction into an AI, if the AI is the one making those like creative decisions as to exactly how it's going to come out, it's not copyrightable. Got it. So perfect segue. Two friends of mine, Fred Royard, a very noted creative director and another great creative director, Chelsea Steiger, they worked together. They prompted one of the, I think it was Dolly Earmid, Jr. or something. They prompted Helma Afklint, you know, the incredible artist, and birds. And the result came out was beautiful and very abstract and interesting, but there was undeniably elements of the artist. And in this case, there is a little bit of nuance with the copyright because as you looked into it, would you explain that a little bit more? Yeah. So Klint died in 1944, which is 78 years ago, and U.S. copyright law protects creative works for 70 years after the death of the artist. So in the case of Klint, those works are no longer protected by copyright and you wouldn't have those sort of like derivative use claims, right? It's the same reason that right as soon as we had 70 years after the death of Van Gogh, you started to see all those immersive experiences because suddenly it's public domain. So those are public domain, but like if you were to do a work in the style of a human artist that's alive today, you might get into sort of infringing territory. Style itself is not copyrightable, so you could do something in the style of the Beatles or you could do something in the style of Taylor Swift or Banksy or whoever. That's fine. But where you get into like possible copyright liability is if the work itself that's been generated includes some component. And so I think it's really important if you're going to be using anything that's generated by AI to know where the source is and to make sure that whatever the company is that's running whatever system you're using has vetted their source material and all of their training data and has put in like certain protocols to make sure that like whatever is generated is not infringing. Makes sense. One of the things in an earlier conversation, Jamie from Joan was talking about using a lot of these ways to kind of mood board, to ideate, like the earlier stages of concepting and creativity. You have, I think she referred to it as like getting past like the blank page syndrome. So there's two uses that I think are very valid for a lot of creatives in this audience and makers in this audience. Using generative AI to mood board or to start showing some directions. Like visually this means, hey, take my brand and stretch it five ways or take this visual direction and explore where we could go. Is that, you know, for internal ideation purposes within an organization, a company, do people need to be concerned with that or is it okay provided it kind of stays in the house? I mean, I think as a copyright lawyer who represents creators, I have to say that any reproduction that's made without authorization is an unauthorized reproduction. When you're working with stuff that's internal and it's not going somewhere and you're not commercializing it, it probably functions very similar to tearing a page out of Vogue or a Pinterest board. So internally I think that's a great use for it and pretty safe. What's interesting is those generative things that you put to kind of generate, they can take things multiple directions much farther than that static mood board, right? You can see variations and hallucinations on the theme in a weird and interesting way. But as we're creative agencies and makers are thinking of the next step out of that, it's very important that the artist, the art director, the copywriter perhaps takes inspiration from that and then builds from that direction, right? Yeah, and it's important too if you want to protect your branding because if AI-generated material is not copyrightable, you also can't stop someone else from using it. So if you have AI-generated art and that's part of your branding, you could get trademark protection for certain things and that doesn't have the human authorship component. But in general, if I'm going to have an AI generate a song or a jingle or something for an ad, I can't stop other people from using that if I don't enjoy copyright protection. So I think there's both the liability side to think about where you don't want to accidentally infringe, but you also want to be able to protect your content. And so that human component is really important and I would think about sort of both sides of that as you're looking to use AI in your business. There's an important distinction because there's a lot of creative agencies that are working on behalf of brands and then there's brands. So if you're sitting within Coke and the trademark and the iconicity there, you have to be very careful as a steward of that, right? Yeah, for sure. The last thing, I wanted to talk about getting a little future-facing here. We were joking, what does... Let's say that there's all of the rights to every Prince recording ever made. And what would it be like allowing someone to sort of API into that for a limited amount of time to create some sort of generative thing, right? Whether it's art, whether it's a new song, is there a role for the people that own these rights to be doing future-facing like limited time licensing in order for a creator, an artist? I mean, imagine what Brian Eno could do if he's tapping into the combined work of something, right? Oh, yeah. And so what gets very interesting to me is ways that are still compensating the estate or the ownership while also allowing an artist a limited dip into something like that. Yeah, for sure. And I think I would not be surprised at all if we see that. I think when you have the rights holders involved in it and consenting to it, I mean, who knows what that could look like? It would probably differ based on whoever's catalog and whatever they were comfortable with, right? But like, or the estate in the case of Prince. But I think that's a very real possibility. And it's also compelling. One of the things I love the most early on with this project that Noah was working on is all of the generative imagined brand collaborations, right? And when something has really, really strong brand codes, it kind of comes through. But so many brands are collaborating. And let's say I'm a small brand that's collaborating with big brand, the big brand could give the small brand access like we would now, hey, you could go into our archives and dig around and find something. But in the future, it might be like, oh, we're going to give you limited access to kind of vacuum what you want to, to see what you come up with. So there's something very interesting about the sanctioned collaborations where rights holders are respected, but it's done with a limited time, limited window and using some technology. I'm using the kind of API reference, but like, who knows what it will be? Yeah. I mean, take the Grimes case. What if someone put out a really cool album and she got to say, yes, I think that use was good and I'm okay with this and I'll get 50% or whatever, you know, the split is. Yeah. I love, Sam Valenti and I had coffee earlier and he cited the Grey album, right? Which was like Danger Mouse and, you know, that set aside. It made a career, kind of like supercharged a career. It had a lot of rates issues with it, but it was like a piece of art that was kind of compiled from a lot of recumbent things. And like the future of that is interesting from an artist respectfully doing something and then it's sanctioned and that it propels a new career. Right. I mean, take the Bob Dylan new basement tapes, right? You could have those in his voice and he could decide, you know, I think there's massive potential there. Yeah. I think that's what's exciting to me is there's obviously huge, scary implications legally and for the rights owners and the respect of original craft and creativity. But I think as we, as we, as this world is built and comes together, there's some interesting territories to like make new stuff, right? Yeah. I think that's where the value is going to be when the novelty wears off a little. Yeah. I think we're in a, we're in a phase right now where it's like the open possibilities, but as it matures a little bit, where does that value lie? Yeah. Cool. Well, thank you for joining us. This was awesome.

BrXnd is coming back to NYC on May 8, 2024 for another full day of marketing and AI—
BrXnd is coming back to NYC on May 8, 2024 for another full day of marketing and AI—
BrXnd is coming back to NYC on May 8, 2024 for another full day of marketing and AI—
BrXnd is coming back to NYC on May 8, 2024 for another full day of marketing and AI—
BrXnd is coming back to NYC on May 8, 2024 for another full day of marketing and AI—
BrXnd is coming back to NYC on May 8, 2024 for another full day of marketing and AI—

This inaugural BrXnd Conference is a day to explore the role of AI in marketing. Learn about the strategic/creative potential of AI for brands—with a special focus on the amazing things that are possible today.

BrXnd.ai is an organization that exists at the intersection between brands and AI. Our mission is to help the world of marketing and AI connect and collaborate. This event will feature world-class marketers and game-changing technologists discussing what's possible today.

The day will include presentations from CMOs from leading brands talking about the effects of AI on their business, demos of the world’s best marketing AI, and conversations about the legal, ethical, and practical challenges the industry faces as it adopts this exciting new technology.

Attendees will return to work the next day with real ideas on how to immediately bring AI into their marketing, brand, strategy, and data work. We hope you will join us!

Who is the BrXnd Conference For?

Marketers

Advertisers

Creatives

Strategists

Executives

Technologists

Data Scientists

Media

We hope that you will join us on May 16th. If you have any questions, please get in touch.

Content

Dispatch

BrXnd Dispatch is a newsletter covering the intersection of brands and AI.

All Posts