3093: AI vs. IP: The Debate Over Creativity, Ownership, and the Law

[00:00:03] Welcome back to the Tech Talks Daily Podcast, where I'm at Web Summit in Lisbon. And today,

[00:00:11] I'm excited for the opportunity to sit down with a true expert in the intersection of law, technology,

[00:00:19] and media. His name is Ed Clarice, and he's the managing partner at Clarice Law. And Ed brings

[00:00:26] with him a wealth of experience from his 17-year tenure at powerhouse brands from ABC and Disney

[00:00:33] to the New Yorker and so many others. But today, we find ourselves at the show floor at Web Summit,

[00:00:40] and also at a pivotal moment in the digital landscape. So my guest today is speaking on a panel titled

[00:00:47] AI vs. IP. He's going to be talking with Dan Milmo of The Guardian and diving into the future of

[00:00:55] copyrightable content in an era where artificial intelligence is no longer just a distant concept,

[00:01:02] but a tool being rapidly embraced by Hollywood studios and media giants. So together, we're

[00:01:10] going to explore the real-world implications of AI on intellectual property rights, and also

[00:01:15] the challenges that it presents both to creators and legal professionals. So what does the future

[00:01:22] hold for copyright law as AI shapes the boundaries of creativity and ownership? Well, let's find out.

[00:01:29] So a massive warm welcome to the show. Can you tell everyone listening a little about who you are

[00:01:35] and what you do? Sure. Thanks for having me, Neil. I'm Edward Clarice. I am a lawyer based out of New

[00:01:41] York where I practice intellectual property as it relates to artificial intelligence and as it just

[00:01:48] relates to itself. I represent media companies, movie television, podcasters, documentary filmmakers,

[00:01:56] digital media companies. Fantastic. Well, it's a pleasure having you on the podcast here at Web Summit.

[00:02:03] And of course, AI-generated content is rapidly gaining traction in media and Hollywood. So from your

[00:02:10] perspective, what are the biggest legal challenges that this technology poses to existing copyright

[00:02:15] frameworks? Because IP and AI is a topic that's increasing, isn't it, at the moment?

[00:02:21] It is. It takes up a lot of my day. There's some huge issues that are happening. More and more

[00:02:27] content's being created through AI. What is created by? It means that there's algorithms that have been

[00:02:32] trained on lots of other pre-existing content, and those algorithms are capable of outputting

[00:02:38] pretty much whatever you'd like. Audio visual content, audio content, textual content, illustrations.

[00:02:46] And so the question is, how does it get created? Training data, for sure, but also prompting by human

[00:02:54] beings. And some prompting can occur through hundreds or even thousands of prompts, and it can happen in

[00:03:01] iterations. So you can have an evolving piece of content that comes out. That, what one might say,

[00:03:07] is a tool where they're using it almost like a PowerPoint presentation or as a computer-generated

[00:03:13] graphics, you know, in CGI. There's a lot of different ways to use the generative AI, and

[00:03:21] people are getting better and better at it. So Hollywood is trying to figure out what it means,

[00:03:27] and there are a lot of legal issues. There's questions around authorship. If the generative AI creates the

[00:03:33] content itself, has it been authored by a human being? A pre-existing requirement in order to have

[00:03:40] authorship. Pretty much every country in the world, whether it's China, Japan, Australia, UK, EU, US,

[00:03:48] all require some human input. But they're all coming out a little bit differently in terms of how much

[00:03:53] human input. Some countries have said that prompting alone is adequate for authorship.

[00:03:59] Others are saying that prompting is not adequate. The US, for example. The US has taken a pretty strict

[00:04:06] rule that generative AI does not have authorship human being and therefore no copyright. Whereas in

[00:04:12] China, for example, and the EU and UK, there's more permissiveness around what prompting means and whether

[00:04:19] there's authorship. And then there's questions around infringement, the training data, and whether the

[00:04:25] training data has been taken without permission and used to create new content. And so we're seeing a

[00:04:33] world where there's a lot of potential litigation and controversy around what AI will mean for Hollywood

[00:04:41] and other content creators.

[00:04:43] Yeah, it is such a big topic right now, especially with the increasing adoption of AI by studios and media

[00:04:50] companies. So with all that in mind, how do you see the role of traditional intellectual property

[00:04:55] laws evolving to meet this new requirement? And are current copyright laws equipped to handle

[00:05:01] the complexities of AI-generated works? Or do you think a legal overhaul is almost necessary?

[00:05:08] Well, we've had anxiety since the early days of technology. It used to be, obviously,

[00:05:14] that everybody would create content through the quill and then the pen and the pencil. And then it went to

[00:05:19] the camera and photography. And then late 1800s, there was a case that went up to the Supreme Court

[00:05:25] of the United States, where people worried that cameras would get in the way of real content creators,

[00:05:32] and that any photograph would sort of commandeer the whole space for visual artists, for example.

[00:05:42] And the Supreme Court said that even though a machine was being used with a camera, it was still

[00:05:48] what the party who was controlling the camera, the machine was the author of the work, and that that would be

[00:05:54] a copyrightable work. So from there, we've seen lots of technology come in the way of content creation.

[00:06:02] There's video cameras, and there's all sorts of technology that's used every day. There's animation

[00:06:09] technology used to be that when you saw Mickey Mouse, every single frame was drawn. Now, the frames are drawn once,

[00:06:18] and then the rest of the frames are created by the computer. Those are well-established,

[00:06:22] well-accepted uses of technology to create content that's copyrightable.

[00:06:29] Do we need an overhaul now that we have AI? Potentially. There is this renewed anxiety

[00:06:35] around whether AI is going to commandeer the space, just like the camera in the late 1800s.

[00:06:42] There is this worry that we will no longer be creating creative content. And then there are

[00:06:50] other people who are thinking that AI is going to be a great partner and will give us the opportunity

[00:06:55] to think of all sorts of new ways to make content that is copyrightable and is enjoyable and is

[00:07:01] ultimately owned by a human being. So I think it really depends on who you are and whether you're

[00:07:07] anxious made by technology or you feel like technology is an opportunity.

[00:07:13] A hundred percent. And of course, there is an ongoing debate about whether AI-generated content

[00:07:19] can be considered authored and thus copyrightable. So where do you stand on this issue and what

[00:07:24] implications does it have for both the creators and the content owners?

[00:07:28] Well, I believe that anybody who gets involved with creating content as a partner with AI should be

[00:07:36] the author. That would be my personal opinion. There are those who believe that more than just

[00:07:43] prompting is needed in order for authorship to accrue to the individual. But I think that if we

[00:07:50] disincentivize people from using the technology, we're going to be losing a great opportunity to have

[00:07:56] this tool which will create new works in conjunction with human beings. Now, certainly there are some

[00:08:04] solutions out there where the content is just generated without any human input and is put out

[00:08:10] immediately into the internet. That's the exception to the rule. Generally, what's going on is that

[00:08:17] everybody in Hollywood and media are using them as tools and therefore they're being more productive.

[00:08:27] They're being a little bit more creative because they have lots of more ideation going on.

[00:08:33] And I think that that's a great opportunity and we shouldn't disincentivize the partnership with

[00:08:38] the technology.

[00:08:39] And of course, we're here talking at Web Summit in Lisbon. And given your extensive experience

[00:08:45] with media giants from ABC and Disney to the New Yorker, I'm curious from everything that you're

[00:08:51] seeing, what shifts are you seeing in how major studios are approaching AI in content production,

[00:08:56] especially regarding rights management, licensing and all those big topics too?

[00:09:01] It's actually a really difficult question. I currently represent over 300 media companies.

[00:09:06] I did work at some of the largest ones in the world and they're having some difficulty because if

[00:09:11] what they create using AI as a tool is not protected by copyright and therefore cannot be monetized

[00:09:18] over a long period of time without falling into the public domain, they're worried about using it.

[00:09:23] So some people are using it and not telling anybody. The copyright office says that you need to certify

[00:09:29] what component of your content was created by AI versus what was created by human input.

[00:09:36] And you're supposed to come clean with the U.S. copyright office. That's going on also in other

[00:09:43] countries as well. So what companies are doing is largely either not talking about what they're using

[00:09:50] or when they are talking about it, they're trying to come up with new solutions for how to license

[00:09:55] their content and monetize it. For example, there's the blockchain. Blockchain is a way to control

[00:10:01] licensing of content without necessarily copyright. It would be through another technology of encryption.

[00:10:07] And I think that this new world of AI and unprotected or public domain content may well be served and be

[00:10:17] solved by new licensing schemes, including through blockchain.

[00:10:24] And I think we need to remind everyone listening that generative AI tools are often trained on

[00:10:28] vast amounts of existing creative content, which raises concerns about unauthorized use.

[00:10:34] We've also heard Sam Altman. I think he was talking about if he's on the internet, it's fair game.

[00:10:39] Apple allegedly accused of scraping data from YouTube, etc. So how should companies navigate

[00:10:46] the legal risks associated with using copyrighted material for training AI training data sets?

[00:10:52] Because at the moment, it seems it's okay, but it almost feels like in the future,

[00:10:55] this one could come back to haunt some of these companies.

[00:10:57] Yeah, well, there's a bunch of lawsuits in the United States and around the world that are

[00:11:01] challenging the scraping and using of other people's copyrighted material to train AI.

[00:11:07] In the US, which is where I focus, there are a couple of cases that courts have already decided

[00:11:14] where it looks like using content in training is not going to be copyright infringement.

[00:11:21] And largely, the fair use doctrine, which we have in the States, is a way for companies to,

[00:11:29] on the one hand, respect IP and the other hand, use IP for new and transformative purposes.

[00:11:36] Whereas in other countries, the fair use doctrine really doesn't apply and strict copyright rules

[00:11:42] do. So we've found that in the United States, at least, there's been a lot of technological

[00:11:47] development. All of the AI companies have pretty much come out of the States. And largely,

[00:11:52] one could say that's because there's permissiveness. And on the other hand, those who want the status

[00:11:58] quo are concerned and they're suing. And they don't want their content taken without their permission

[00:12:04] and without payment. There are some situations where the AI companies are now licensing and paying

[00:12:09] a lot of money to companies. There was this word that Fox got $250 million for a license to give over

[00:12:16] their audiovisual content. But many others are not being paid and are suing, including the New York

[00:12:22] Times, for example, which has a huge lawsuit against OpenAI and Microsoft for illicitly taking their content.

[00:12:29] So we can see that there's a multiplicity of ways to go about this. And the courts in the US and other

[00:12:35] countries will have to solve this very sticky problem, protecting human input and human authorship,

[00:12:42] but at the same time endorsing technology.

[00:12:45] For many media and entertainment companies, the promise of AI is balanced by fears of

[00:12:50] legal disputes in the future, especially over ownership and copyright infringement. So

[00:12:55] for anyone listening anywhere in the world, what would you say are the best practices that you'd

[00:12:59] recommend to businesses, maybe looking to leverage AI while also mitigating potential legal

[00:13:04] pitfalls in the future?

[00:13:06] Good question. I oftentimes give advice to my clients that they should use AI for sure as a tool.

[00:13:12] They don't need to advertise that they're using it. It's another piece of technology. We all use

[00:13:17] Microsoft and we use all sorts of tools and we don't necessarily advertise that. We just have it help us

[00:13:25] create as a partner, as a tool. And I think that people need to get good at it,

[00:13:32] prompting the art and science of prompting is something that comes with time. And I find that

[00:13:38] even in my legal practice, I'm practicing using the AI tools in order to help me with some

[00:13:45] brainstorming on solutions. Oftentimes what I'll do is I will come up with an idea, write out my whole

[00:13:52] strategy for a client. It could be many different parts. And then what I'll do is I'll have the AI

[00:13:58] answer the same kind of question that I've already done just to see if I've missed anything.

[00:14:02] And also to see what my client might be looking for in the AI. There may be a day when they're

[00:14:08] better at prompting an AI to give them legal advice than they are at coming to me. So I want

[00:14:14] to make sure that I'm anticipating what the AI is going to tell them.

[00:14:18] And as businesses increasingly turn to AI for things like code generation, how important is

[00:14:25] private AI? And do you think that resolves any potential IP implications from integrating AI

[00:14:31] generated code within their proprietary products and applications? Because this is a question

[00:14:36] that I'm getting asked more and more lately.

[00:14:38] Yeah, that's a good question. There are two different ways that you look at training data. One

[00:14:44] is through a large language model where there's a huge amount of content coming in from many different

[00:14:50] sources, as we said, scraped from the world. And then there's also more proprietary AI

[00:14:57] language models that are just trained on things like your own emails and your own documents and

[00:15:03] internal corporate material that you're providing to the AI in a closed environment. And places like

[00:15:11] Microsoft have come out with a product called Pilot. And they're making some promises that the content

[00:15:20] that you put into their solution when it's on the private side is secure and not going outside the

[00:15:26] firewall. I think that that's probably something that we can rely on. Microsoft is a very good company,

[00:15:33] and they're pretty good at security. But be careful, you know, because you don't know who you're

[00:15:39] partnering with. And if you are providing content that could potentially get out, that would be a

[00:15:45] breach of your own client's privacy or your own privacy. And the lack of security would be really

[00:15:53] problematic.

[00:15:54] And you're someone who teaches media law and intellectual property. So I've got to ask,

[00:15:59] how are you preparing the next generation of lawyers to tackle some of the unique challenges

[00:16:04] that AI presents in these fields? And are there any key legal precedents or cases that you think will

[00:16:10] possibly shape the future of AI and copyright?

[00:16:13] Well, my students are all probably ahead of me in terms of their ability to play with the tools.

[00:16:19] They're all using the tools to help them write papers and come up with ideas. I have young

[00:16:26] associates who also use AI to help them be better lawyers. So largely, it's making sure that they

[00:16:33] understand the base principles of copyright and IP, making sure that there is an appreciation for

[00:16:41] what is owned and what is shared, and understanding how technology has been used by courts as a way to

[00:16:49] both advance the law and also protect against inappropriate behavior. And so I oftentimes look at

[00:16:56] balancing different interests with my students and have them be sensitized to both points of view,

[00:17:03] and I'm going to help them out. And I'm going to help them out. And I'm going to help them out.

[00:17:07] Well, I know how incredibly busy you are. And you're going on stage tomorrow, I believe,

[00:17:11] here at Web Summit. But for anyone listening, maybe they want to reach out and find out more information

[00:17:17] about the kind of work that you're doing. Dig a little bit deeper on this topic. Is there any way

[00:17:21] you'd like to point everyone listening?

[00:17:22] Sure. Edward Clariss. I'm on LinkedIn. Always reach out there. I'm on LinkedIn often. My law firm is

[00:17:30] clarisslaw.com. And come visit, get in touch. Happy to talk.

[00:17:37] Well, I'll add links to everything so people can find you nice and easy. But thank you for stopping by,

[00:17:42] and best of luck in your keynote tomorrow.

[00:17:43] Thank you, Neil.

[00:17:44] I think it's clear from our conversation today with Ed that the rapid adoption of AI and media

[00:17:49] and entertainment isn't just about reshaping business models. It's also about challenging

[00:17:54] the very fabric of intellectual property law. And Ed's insights highlight both the opportunities

[00:18:01] and the legal complexities that we need to navigate as AI-generated content becomes the new norm.

[00:18:08] And as we look ahead, the key question remains, how will the legal system evolve to protect creativity

[00:18:15] while embracing innovation? So much food for thought there, here on the show floor at Web Summit in

[00:18:22] Lisbon. But wherever you are listening to this conversation today, wherever you are in the world,

[00:18:28] let me know your thoughts. Please email me, techblogwriteroutlook.com, LinkedIn, Instagram,

[00:18:34] and Twitter, X, just at Neil C. Hughes. Let me know your thoughts. And I'll be back again tomorrow

[00:18:39] with another conversation here at Web Summit. But enough from me. Time for me to look for some

[00:18:45] free tech swag. I will speak with you all again tomorrow. Bye for now.