Carlo Rizzo 00:00:50

Good morning, everybody. Thank you so much for joining us. And we are here with our amazing panel about the future of IP. We are presenting a publication called "The Future of IP in the Age of Machines." I am Carlo Rizzo. I'm the co-founder of CinemaO and the editor of the Black Cube, which is the research platform that has undertaken this research. What I'm going to do is that I'm going to briefly introduce the panel. Then we will go through a little bit of the findings of the research, and then we'll go on to our discussion. I'll try to be as quick as possible. But here we have Angela Dunning, lawyer from the US based in California, partner at Cleary Gottlieb. And Yannick Bossenmeyer, who's the CEO of Cascade8 and co-founder of Logical Pictures. Erik Svilichh, who is the founder and CEO of Encypher. Hi. Elisa Alvarez, who is the managing partner for global intellectual property at Tower Peak Partners. So I'm very delighted to be here, and it's an amazing array of perspectives on what the future of intellectual property is going to be, from an investor, a technologist, a hybrid producer, technology and investors, and a legal expert and scholar and... And lawyer. Practicing attorney. So what I'm gonna do is that I'm just gonna briefly give you the context. This project started about 6 months ago. We really wanted to, we were really asking some questions and about what's gonna happen with machines generating content. And the starting point was a conversation with Elisa and Elisa's team. Yeah. In October, in which we realized that we were asking ourselves the same question. And that triggered the... this idea of going deep and interviewing industry players at the frontiers of... at the frontier of IP development and sharing and valuing, and really try to take also a value perspective, a value chain perspective, and an investor's perspective. And then we were lucky to have Erik coming on board, and also supporting the research with being, Erik, someone who within Encypher has a really deep interest and influence in how we are establishing a provenance infrastructure. So we got together and we, I don't know if this sounds like a large or a small number to you, but one of the things we realized is that everything we figure out now, everything we found out now is going to be obsolete within a matter of 6 months. So we... Yeah. So we try to find a balance and we interviewed 28 people within a space of 2 months. We went to print last Sunday and then we arrived with this one in Cannes. I think speed was of the essence because of the nature of the subject. But we try to cover as much as possible the span of the industry, bringing lawyers, of course, in the forefront, but also technologists and founders and producers and industry experts, all being interviewed and asked deep questions about what is the future of IP from a value chain perspective. So what I'm going to do is that I'm going to give you a very brief overview of some of the findings, but then we'll go deeper into the conversation with our panel. And the publication is going to be available today for free for everyone on our website, which I'll show you at the end. And we also have it in print, and you should be... Thank you. Receiving at some point some... a quick summary of it as well that will allow you to also take a look at the content. But so the first thing, there are new risks emerging and of course new solutions. Attribution, the very pillar of IP development, and is obviously... there's new uncertainty generated by AI AI-assisted production. The legislation is still adapting, as we will find out with, with Angela, and the industry is self-regulating. And there are newer insurance products that are emerging, but they tend to be quite expensive. And it's an interesting space to look at with insurance tech. Providence enforcement is a new mantra, I think, right now, and not just because Erik is here, but because it's really important. There is a new infrastructure that has already been developed. In fact, This is incorrect. I say a new infrastructure is needed. A new infrastructure exists. It needs to be adopted. And to enable an effective transparency layer, we have watermarking, we have fingerprinting, we have various independent verification technology that are gaining adoption. This is a really important space to look at. We are also facing a transparency dilemma. You probably have heard multiple times on stages here in Cannes, someone telling you, make sure you record... if you're a producer, if you're an author, make sure you know and you track how you're using artificial intelligence. But then that opens up a question: what happens if you track everything to the creative process? Now the creative process is effectively fully monitored and fully registered. Is that good for the industry? Do we have a dilemma here looking at transparency but also scrutiny? And this is an important question that was asked in one of the interviews. We have a technology that is being deployed across different different elements of the sector, different points in the value chain, but licensing frameworks are not catching up yet. And I think this is an important... I would like to see this as an opportunity for licensing frameworks to adapt. And I think there is a real strong case there. And again, also something that probably you have heard a million times, but especially in the EU, tech and data sovereignty is a really... remains a really important issue. We will talk about that with Yannick as well, who has published recently an amazing white paper on new AI tools and how they're being used in the industry, which you'll tell us more about. But that's another important topic here in the US and one of the key cross-border issue vis-à-vis the... here in Europe and one of the key cross-border issue vis-à-vis the US and China especially. There's another interesting fact emerging, which is that There is a lot of blanket AI restrictions being put either on talent or tools being used because we don't know how these are going to be adopted. We don't know how to manage the risk. And this is actually bringing us to a point where talent and resources might be underutilized, and we should be looking at how we develop greater transparency and certainty to make the most of the talent and the tools that we have. But there are also great opportunities that are, I think, are important mentioning and looking into. Again, everybody talks about IP as an ecosystem these days, but what does that mean in practice? It means that the source assets are really key. And it's important, especially from an investor's perspective, to look at these source assets and what potentials they have. And again, Also the question of what is really truly the impact of AI tools on IP development, and probably the most important impact is that we are entering an era of rapid prototyping for film products and audiovisual products where the cost of failure comes down. It's not about changing and transforming as much the high-value end product, but it's about getting more new products in the market that can be tested more easily and sort of shorten the time to market, which in in our industry tends to be quite, quite long. Another really important opportunity, which is that when thinking about investing in a piece of IP, data and the data that is connected to this IP now becomes an asset which is almost as important as the idea. And it's not about necessarily buying data together with IP rights, but it's about buying access to that data. And it's very... it's going to be very important going forward what data comes with your IP asset. And also, if you are a creator or a producer, what data you have access to that you can pitch to your funders or to distributors that you will be able to share and you will be able to package as a way of increasing the value of your asset. Then, of course, there is participatory IP and the way in which fan engagement we have here. Phil McKenzie was one of the people that we interviewed and who's developing a new product in this respect is really...

Audience member 00:09:54

Yeah.

Carlo Rizzo 00:09:55

The future prospects here is a huge market and there is a strong demand from fans to be part of IP development and this is an area that we should be watching. There is a broader area also that is important to look at, which is we talk about chain of titles, but really now titles are becoming networks and we are seeing audiences becoming creators, becoming investors, and we are seeing talents becoming creators, becoming investors, and becoming IP holders, and this intersection of roles means that there isn't a linear chain of data anymore out there, and we should be looking at interoperability of data formats and channels more and more. And there are important initiatives in this respect. There is a company called ValueNode, which I would recommend you look at into... and the interview in our, in our publication, which is also doing this work at the EU level. And it's very important work, fully integrated within Encypher as well. And something that I really care about, which I think we shouldn't forget, all our discussions tend to be about the digital world and where we are in the digital world and the sort of ephemeral assets. But the reality is that the more we work and we see ephemeral assets, the more these become part of our life, the more there is going to be demand from people who are producing high-value assets to also store them and protect them in physical formats. Yeah. We interviewed the founders of the Arctic World Archive, which is almost like a seed bank. In fact, it's next to the seed bank in Svalbard, and it's protecting with a physical infrastructure under the permafrost of Svalbard the most precious cultural IP in an immutable format that could last 2,000 years. It sounds like a wild thing, but the reality is that... It is. A demand for preservation for physical IP assets is emerging, and I think it's very important to remember that we don't just live in a digital world. So I'll finish this. The full research is available in print, on our website. Please email us if you have any questions, but now we should go back to our amazing panel. So thank you all for... for being here. And again, thank you, Elisa and Erik, for your support throughout the process of bringing this, and thank you, Yannick and Angela, for being here, and Angela, especially for you, for traveling all the way from the US to be part of this. We are very, very grateful. And perhaps, Elisa, I'll start with you, if that's okay, because we started with this investor's perspective, so we really want to know From your perspective, what is the outlook of the IP market and what is most exciting that you see ahead?

Elisa Alvarez 00:12:53

As a firm, as a private equity firm backed by institutional investors, we are very bullish about the potential for intellectual property, particularly in the entertainment and sports space. So this is where we are focusing at the moment. And the main reason for the firm to have taken the decision to bring IP and entertainment in particular into one of our areas of focus is the, is the nature of the IP as an asset. The ephemeral, as you, as you mentioned, nature, how differentiated it is in the market. But primarily, it fits into our own ethos as a firm to bring value to investors through focusing on industries in transition with high barriers of entry and is perceived at the time when we come in as a niche. So IP fits into this framework perfectly. And we see enormous opportunity. And me personally, having been in entertainment for now well over 20 years, 25 years, we do see that the cycles come and go, and right now, thanks to, as it has been in previous cycles, technology has informed the changes there, the very structural changes, often, to economic models that the space has gone through in the past, except that now, it's accelerated by unprecedented, use of technology. Of course, with the instability and the new economic models come uncertainty, which is not a feature that most investors deal with comfortably. We like to say that we are comfortable with instability. We know how to navigate those cycles. There's... there's room for great expansion and great achievements from, as we see them, both from creative and commercial perspectives. And we as investors are excited to be supporting those changes.

Carlo Rizzo 00:15:50

Thank you, Lisa. And maybe Yannick, I wonder whether you have this multiple perspectives, both because you are a producer, but you also invest in other productions and you run a technology company. I'm curious to see also what your perspective is on where we are going with IP development in the coming years.

Yannick Bossenmeyer 00:16:15

Yes, Logical has a specific position in the field because we... well, we come from finance. So our first job is to finance movies. But now we produce, we sell. Both locally and internationally. And we have this tech subsidiary, Cascade8. And so we have this trend to see, to try to innovate in what we do. So of course we love tech. We worked with blockchain, with Web3, and for a few years about AI. We use AI in different ways. Analytical AI, it's not such a subject here, but it's important in the field. We use it, for instance, for Blockframes, that is a tool, software, to manage film revenues. And we use AI to read the film contracts and to transform them in an algorithm and to automate the waterfall construction. But there is no IP subject here. The IP is more, of course, on the creative side. And I really see two trends in the company. The AI lab indeed is an advertising subsidiary called Love Boat. They are really willing to use it and they use it already. They produced an ad for Salomon, Shaping Your Future, that is 100%... well, not 100%, 80% AI-generated. They are doing also an AI ad for Peugeot that was just out. And on the other way, there are the feature film producers. It's less easy to convert them to AI. They are more afraid about about IP preservation, about the jobs preservation, actor technicians and so on. But my belief, and I guess the belief that everyone here, is that AI will transform things but not destroy the industry.

Carlo Rizzo 00:18:32

Thank you, Yannick. And speaking about transformation and the need for protection, Erik, Maybe you want to tell us a little bit about what does the provenance infrastructure of the future look like, because we are in a space where this is becoming more and more important, especially in the context of what Yannick was talking about. And maybe it's worth explaining a little bit what Encypher does and what does it mean for the industry.

Erik Svilich 00:19:04

Yeah, so, What we do at Encypher is essentially is an API that embeds cryptographic metadata into images, video, audio, and text. We also do watermarking and fingerprinting. So essentially, it's also based off of a global open standard called the Coalition for Content Provenance and Authenticity, backed by Google, Meta, BBC, OpenAI. And we authored the tech specification for that standard. To be able to embed cryptographic proof of origin into digital assets to essentially prove that your organization has produced a digital asset. It's been edited, maybe AI was used in some way to be able to prove and show audiences that you used AI to modify lighting on a scene, for example, which is very pertinent to the EU AI Act, for example, which is now, you know, a requirement by law. Also in several other jurisdictions like California, it's coming into play, in China it's already in play. India, it's already in play. So it has a lot of different applications. One, being able to prove that the asset that you're putting out into the world was created by you or owned by you, has license and rights associated with that actual content. So when it's copy and pasted, or it's scraped and put in a third-party database, or, you know, uploaded and downloaded somewhere, your license and rights for your asset that you're putting out in the world actually follow the asset itself, which... in a machine-readable way, which is critical in the age where, you know, information is shared so freely and can go into third-party databases, and the license aspect and the ownership aspect of that digital asset frequently gets completely separated from the asset itself. But with this methodology, it's cryptographically attached and verifiable. In a very free and open way to be able to verify those licenses and rights that are associated with that content. So, yeah, it's kind of like the underlying layer of being able to monetize your IP in a way, because if you can't prove that you made something or that it's yours, how do you monetize it? You know, it's very difficult to do so, because somebody might take it and claim that it's theirs, or somebody might make a... A copy. Content very similar to your content and claim that it's yours, but it's actually not your legitimate content. So you need a way of being able to prove not only that this is your content, it's authentic to you, but if somebody comes out and pretends to make, you know, a film trailer about your film, but it's not from you, you need a way of disproving that. Or for like deepfake in news media, for example.

Carlo Rizzo 00:21:50

I'll let you answer this question.

Angela Dunning 00:21:52

Well, I haven't... I'll ask a question. So a hammer always finds a nail. A copyright lawyer always finds a copyright question. But how will you validate and verify that the owner who applies for the watermarking is in fact the copyright owner and did not themselves infringe someone else's work?

Erik Svilich 00:22:12

Great question. Ideally, when somebody is creating a new piece of work and they're relying on external assets, those assets would have provenance in my perfect, maybe utopian/dystopian world that we may live in, in the near future. So essentially, from our perspective, we work kind of like a notary, and essentially, we will get an audit trail of what went into the creation of that asset, and embed that as part of the metadata that goes into that. So we're relying on their representation of, of that creation of that asset. And we're not saying, and necessarily 100% verifying, that we know that 1,000 edits went into this piece of work, and every single one was done by a human creator, partly AI. You know, we have a way of marking and telling what's AI and what's not, if they're a good actor, you know? But if...

Elisa Alvarez 00:23:16

Yeah.

Erik Svilich 00:23:16

There's always gonna be bad actors and people trying to use AI and maybe pass it off as their own original creation. You know, it's very difficult to prove that if somebody's very determined, right, to remove all that information. So we're not lawyers, and we're not asserting that they have absolute copyrights. We're letting the asset owner assert that themselves, and we're the transport layer for that, to be able to embed and verify that information.

Carlo Rizzo 00:23:48

Thank you. Angela, let's continue with you. You had a... you've been focusing very much on artificial intelligence and working and representing LLM providers and developers. And you played a really important role in establishing what is known as the fair use sort of framework in the US for training data. So I'd like to know from your perspective, what your reactions are to some of these conversations, and what do you see as the most important developments in the legislations we will see happening soon that will affect how we create and develop intellectual property?

Angela Dunning 00:24:33

Thank you so much. So I share a lot of the viewpoints on the stage. I'm both excited about the technology to come. I'm cognizant of the issues we need to work through in terms of how we figure out how to navigate the new world. For better or worse, a lot of the law is being developed in the United States where I practice. And I do represent just a huge swath of major AI developers and investors, as well as content creators and very large platforms. And within my clients' organizations, there are often different and sometimes conflicting viewpoints. We want to use as much data as possible. We want to protect our...

Carlo Rizzo 00:25:14

Our data.

Angela Dunning 00:25:15

IP as much as possible from use by others. How do we navigate all of that? In the United States, we start with a few propositions which are fundamentally different from Europe and are worth understanding. One is that no one can own an idea or a concept or a style. US law is based on a premise that we want to encourage new expression, and that means we allow people to build on and critique and... And innovate. Expand on the work of others where they convey a new message. That's a sort of combination of our fair use doctrine, but also first principles of copyright. And so in the United States, the big issue we've been litigating over for the last few years, and I've been handling AI litigation since the first case was filed, is this fair use question, right? For most developers, there's, there's no dispute that they copied copyrighted books or articles or song lyrics or other content from the internet to train their models. The question is whether they have put those copies to what we call a transformative purpose that does not harm the market for the original, but creates new expression. And I worked on a case for Meta, in which we have gotten to judgment in the case, and the court did find that Meta's use of copyrighted books to train its models was fair use. The court found that this was spectacularly transformative, that it enables an entire array of new content to be created. There were no allegations that anything coming out of the model looked like something that it was trained on, and there were no allegations or facts brought forward that plaintiffs had sold any fewer books or lost any revenue because of this training. And so those... that combination of facts led the court to find that it was fair use. We had a similar ruling in the case that was filed against Anthropic and the Claude model, and the court found that that was fair use. The two courts differed on whether it matters where that data comes from. In the case of Meta, the court found it didn't. Meta put the data to a fair use, it created something new, and that was fair. In the case of Anthropic, the court found that because it had downloaded content from certain websites that it called pirate websites, that that could not constitute a fair use at the judgment stage, which set up a potential for liability in the many tens of billions of dollars, and ultimately led to a settlement that Anthropic entered across the class of about 1 point... it was $1.5 billion. That settlement was just before the court for final approval last week, and we are expecting final approval this week, which means those book authors, there are about 450,000, will receive about $3,100 per work for use of their books to train Claude, but not for the training part. It's for the way that Anthropic downloaded the works. At the same time that many, many cases across the ecosystem remain to be decided, OpenAI, NVIDIA, Apple, Google, the list goes on and on, we are seeing markets develop, as Carlos said, for access to data. So it's not licenses for copyright. I'm... Spanish. Paying you to use your copyrighted work, the developers would say that that is fair use and permission isn't required. And yet we're seeing huge data deals being done because where you can find access to huge volumes of data, the volumes needed for effective AI training, and where that data is clean and it has the metadata required and it is in the format that's necessary, for training, that is a tremendous service that companies will pay dearly for and are happy to pay for. And so we're seeing that develop, that market, at the same time that we are actively litigating the fair use questions and expect to be doing that through appeals for possibly the next 7 to 10 years.

Carlo Rizzo 00:29:37

Thank you, Angela, and I think this goes also back to the... at a very practical level, and Yannick, I want to come back to you also, it comes back to the point of how do we use the tools that we use, what do we know about the tools that we use, what do we know about the data that is behind the tools that we use, and you just published a white paper with Cascade8 that went into quite a lot of depth into some of the most popular tools that are used in production, and you also assess them in terms of their long-term viability and sort of also made a good argument about what to look at when you don't know the... where the data resides and where the contractual aspects of the tools will take you in terms of jurisdictions.

Angela Dunning 00:30:31

So... Yes.

Carlo Rizzo 00:30:32

Can you tell us a little bit more about your findings and what does it mean? And I guess what we're trying to figure out is really how do we create value now. So it's in that perspective that I'm curious to know what you found out.

Yannick Bossenmeyer 00:30:47

Yeah, well, this paper, we called it "A Survivor Guide for European Producers Facing US AI Domination" because... Sorry, Angela. But most tools we use are Americans and driven by American data and our European... well, you talk about value. The question is, for who are we creating value? And when I'm entering my European data in a US tool that will add them to other US data and give an output that is more US-oriented than Europe-oriented, it will fit US platforms or US broadcasters and not the European market. So, I think that these tools are really useful and it's not a matter of not using them, but the object of the guide is just to show where we have to be careful. Okay. When we use them to know the bias. And besides that, there is a point of sovereignty because talking about soft power, the use of this tool is driving America again first before Europe. And we would need to develop a European ecosystem. China has the same, the same situation. Thank you. Thinking, and they managed to do it in the Chinese way, so with constraints and forcing companies to do certain things only with Chinese data and not American data and so on. The other thing is that they have also an enormous market, more than 1 billion people speaking the same language with the same culture. We don't have that in Europe. Yeah. So we have a lot of constraint in Europe, but I think that we have to find an alternative for software. And I don't know, in case, we know that now that it's not fantasy to imagine hard tensions between America and Europe. So what happens if one day US say, well, you don't have access anymore to the US tools, to the AI US tools? We have to have an alternative.

Audience member 00:33:10

Thank you.

Angela Dunning 00:33:11

Angela. Well, I share your concern because one of the things American companies are confronting is the fact that European law with respect to models is potentially fundamentally at odds with U.S. fair use law. So, a company like Meta that released open source its LLaMA models should theoretically give every European company an opportunity for free to use a pre-trained model and develop their own systems to serve European needs or what have you. The problem is that the EU AI Act purports to impose its restrictions on any model whose outputs could enter Europe, whether it was trained there or developed there. And so I think, and there is a rise of... powerful AI development in China, which is not constrained by the same copyright protections. They also used, for instance, LLaMA to develop DeepSeq, and they are operationalizing that at a faster clip. And so, in some ways, Europe and European regulation has constrained what can be developed here in furtherance of the European economy, which, I think should be concerning to everyone. We should want a strong economic sector here for AI, for creators, for what is possible in the art and science, et cetera. So I share your concern.

Yannick Bossenmeyer 00:34:46

Yeah, you're right. And the problem is that Europe is a good student towards moral right and talent protection, but at the end, we are going to pay it as a strong price.

Carlo Rizzo 00:34:57

And the question about China is also very important because there's not only American tools now, there's also a lot of Chinese tools that are available and that are very powerful. And Elisa, I want to go back to you. You mentioned before a very important thing about this idea of navigating industries in transition. And if we have to think about creating value, from your perspective, what does it mean to create value in industries in transition, as we just seen?

Elisa Alvarez 00:35:29

So, I'll speak to my own domain within the firm, which is entertainment and sports. So, we are neither creators of content nor operators in the tech space. We are catalysts, and that's how we, create value and facilitate the expansion and scaling of value that has been created by the entrepreneurs with whom we work. So, in a way, we remain at the margins of the bigger discussion, and we are not the key actors. However...

Audience member 00:36:15

However?

Elisa Alvarez 00:36:16

In order for us to fulfill our mandate and invest with confidence, we do need the tools that the law and the technology may provide us. At the moment, the somewhat uncertainty, not so much in terms of the law itself, because the principles are there, but how it's going to be applied and how the UK, where we seem to have a self-regulated approach at the moment. So it ends up in our hands very much to decide how we're going to approach that. And our take on the technology that is available is one where now more than ever, our chief technology officer, contributes with his and his team's views over investment opportunities. I never thought I would have that. But we bring them in, they ask questions, they participate in the due diligence. I mean, it's creating a whole new... [LAUGHTER] ...this kind of area of work for them. But we feel that that is necessary. On the one hand, yes, there's the protection of copyright, and ultimately the IP is at the center of everything we invest in, is the core. But beyond that, there's other aspects that influence IP, not least because we have a strong focus on...

Yannick Bossenmeyer 00:38:05

The environment.

Elisa Alvarez 00:38:05

The audiences that are created, which in themselves are not an asset, but how you connect with those audiences and how you assess their engagement and value very much depends on the technology that is available. And beyond that, as if I can see circles, right, around the intellectual property, is actually... The data.... and that is a change from a commercial perspective in our approach... the fact that we will invest in brands, right? So it's the IP, it's the relationship with audience, and surrounding it all is the brand. So there's aspects of intellectual property that are influenced in technology. And ultimately, in this somewhat uncertain market, we feel that we need to take control and diligence all aspects of technology that involve the assets.

Carlo Rizzo 00:39:07

And the point we made earlier about data that I was mentioning in the beginning and also Angela reinforced, I'm curious to know from your perspective, does that resonate? Do you see that in your due diligence now? Do you see that as becoming a more and more important aspect of how you value an asset.

Elisa Alvarez 00:39:26

Very much so. So when we value the assets, we look at... or we used to, generally speaking, look at creative aspects, commercial, and financial. And now one of the pillars of the commercial due diligence really is what data is available to support that one asset, the potential audience, and the relationship between the operator. So the... because we tend to invest in or with distributors of content in different permutations. And data that shows us their relationship with their market.

Carlo Rizzo 00:40:12

And one important sort of clarification, I guess, is also to say that Your perspective as a private equity investor is on the slate rather than on the individual title, so you're investing on groups of titles.

Elisa Alvarez 00:40:30

Yes, and we actually have two points of entry. We can invest in the companies themselves, which we will do in a more selective way because it has to be strategic, but primarily we invest invest in the IP assets, however, never as individual IP, always as mini-portfolios as we...

Carlo Rizzo 00:40:55

Yannick, on the other hand, if I ask you to put your investor hat on, you invest in titles. I'm wondering from your perspective also whether this question around access to data about the titles that you invest has come into consideration when at the time of investment and looking into new assets?

Yannick Bossenmeyer 00:41:20

Not really at the moment. We are focused on, well, the artistic and commercial side of the movie.

Carlo Rizzo 00:41:29

This is interesting as something that... I'll be interested, curious to see how this will evolve. And Erik, also from your perspective, You mentioned in the past, when we were discussing and preparing for this, also this idea that provenance is a key to monetization. Can you expand on that, since we're talking about how do we create value, how does that also apply to your work?

Erik Svilich 00:41:58

Yeah, I mean, provenance can be used offensively and defensively for, you know, IP owners. So both for proving that they created this asset is useful for multiple reasons, right? It's when you are putting that onto the web and brands are putting their ad next to your content or in your content, they want verified inventory, that this is legitimate inventory, comes from the original creator. And we're actually seeing brands willing to pay extra to place advertisements next to verified inventory. And I think there's gonna become a new paradigm where if your inventory isn't verified from your organization in some way, um, then, uh, you'll get less of that pie. Or it might eventually vanish if you can't verify your inventory at all. Um, so... but in the meantime, especially for early adopters, we're seeing an increase in willingness to pay for brands especially to place their brand reputation next to or inside of your content, which has been very interesting for us. I lost my train of thought.

Carlo Rizzo 00:43:15

No, no, you answered the question. But I think it's interesting to see all of these different avenues for value creation that are emerging. It's interesting to see the difference between what a more institutional investor would look at versus one sort of direct, more direct investor would look at. And, but this idea of provenance I think also connects to data because it's about transparency, it's about what information are available around a piece of intellectual property, no?

Erik Svilich 00:43:48

Yeah, I would say, on that, the more that you can prove the creative process that went into it, you know, like Angela was alluding to earlier, the more valuable your asset becomes, especially in this era where you have regulations like the EU AI Act, where you have to be able to prove, like, potentially in a court of law, what part of your content was generated using AI, whether that was fully or assisted or edited or whatnot. Mm-hmm. And so those that are, you know, looking to invest in your film, if you don't have the ability to do this, and then it becomes a regulatory problem, could be an issue not just for yourself, but for your investor as well, if you don't have that capability.

Carlo Rizzo 00:44:30

But that is the defensive position. Yes, that's the defensive. The offensive opportunistic position, I think, is also an interesting one to explore. And Angela, we spoke about this idea. It was a speculation, I don't even know if it, I don't know if it's true, if it will happen, but I'd love to know your view of whether a new market for data is emerging almost as if... because obviously the reaction is going to... is a protective one is, you know, we talk about insurance products, we talk about provenance infrastructure, but then at the same time, you mentioned earlier, there are data deals happening and it's possible then even in... without having to go and look large language models that are being trained on masses of data, even if we think about producers that are trying to create their own safe clusters to create their work without having to worry too much about sort of IP infringement risk, maybe there is a new market emerging. I love your view on that.

Angela Dunning 00:45:35

There definitely is a new market emerging for data and for new approaches to training, but If I can sort of pan back for a moment, I mean, I come to this stage not just as a lawyer who's been litigating over brands and trademarks and copyrights for more than 25 years, but as a lover and connoisseur of film and art. I'm married to a writer. I mean, I think we live in an extraordinarily exciting time. These are new tools of creation. They are new powerful tools of creation in the hands of human creators. And there is a way to navigate this new world that allows us to continue telling our stories, but gives us new ways to do that. And there's a way to protect that, and to own it, and to promote it, and to make money off of it at every level of the value chain. At the most basic level, you know, I... again, I come from the US, so I focus on US copyright, but my... our firm is located throughout Europe, and we have clients here and everywhere. In the US, the US Copyright Office was touting just a couple of weeks ago that they have registered thousands of works that used AI to augment or underlie the creation of new works, works of visual art, film, you know, Film has so many different layers of copyright in the script, in the score, in the actual shooting of the film. And there are ways to use these tools that don't risk human authorship. I litigated a big case years ago where a monkey sued for copyright infringement. And the big takeaway in the US was a monkey can't own a copyright. Also, don't lose against a monkey in court. That's career-limiting. I didn't. I'm still here, but human authorship is required. So what that means is AI cannot replace humans. It cannot replace human creators. AI is but a tool that human creators need to make. So you should use these tools. You should document to some extent. You should try to know what... where the data is coming from. You should avoid creating things that infringe the works of others. In the outputs or the final production, because that would encroach upon rights that we've all had for a very long time. But I think at the end of the day, if we approach this as a new challenge, but actually one that enables a really exciting explosion of potential new forms of creativity, we will find the value. Thank you. In every level of the chain. We will find ways to make money because I don't think humans are done telling stories and sharing their experiences anytime soon.

Elisa Alvarez 00:48:32

And the value will ultimately be dictated by whether or not there are audiences, fandom, that once having access, of course we need to promote the access, curate the access even more, But that's where we'll see value or not. I think the ability to establish copyright ownership is a condition for the monetization. So yet another tool, certainly from our perspective, a complicated and very important tool nevertheless, but a tool. To allow us to commercially exploit it, so monetize it.

Erik Svilich 00:49:21

And one more quick point on the offensive or monetization of provenance is, at least in Europe, new models that are being created need to be able to verify the provenance of their training material. So if you're creating new assets and creative works, however they may be created, And you have that provenance chain, that is now inventory for the creation of new, potentially European creative models. Where if you don't have the ability to do that, that's not a monetization avenue for yourself, right? But if you do have that, maybe you can help create Europe's new digital model, right, by verifying your inventory.

Carlo Rizzo 00:50:04

I would like to... thank you, everybody. I would like to leave a little bit of space for questions. We have a few minutes if you can keep them concise. And the lady at the back was the first. Maybe I'll just... is there a roving mic somewhere? Oh yes, thank you.

Audience member 00:50:25

Thank you. Hello, I'm a US lawyer working in films. But I wanted to ask you, the word data has been used, I think, in a number of different ways when you're talking about You're talking about the value of data and licensing data, and... but then you're talking about training on the basis of data. So it seems like you're talking about data to mean, like, copyright works that are used to train and create new works, but you're also talking... and the licensing of that. But sometimes you're also talking about audience data. The information that copyright owners and producers and so on are trying to obtain often from streamers and so on, because the value of their own data as to who consumes their products and so on is very important and is also saleable. Could you just clarify the way you're using data?

Angela Dunning 00:51:23

It's a big tent word. Information, metrics, Audience scores. It's all data at the end of the day, right? It's all information. It's what you do with it.

Erik Svilich 00:51:36

Unfortunately or fortunately, you have to have data with a qualifier, you know, in that conversation, in that context, because it is such a broad term, so.

Audience member 00:51:48

Thank you. Hello. Thank you very much for this session. It's very helpful. It's also quite But as a work in the area of tech and also in creativity, it also feels like a very heavy burden on the shoulders starting to settle. But one of the things in all of the talks that have been to us, we talk about AI, we talk about generative AI, but nobody's actually been able to define actually what we're copywriting in a sense. So there's...... the thing where we teach, we train people in how to use AI for filmmaking, but one of the things is, where does... AI is not just a tool in the usual way, in that it is a tool that can give you feedback. It's got autonomy. There are things that it does. So it's a tool like we've never had before, in that sense. So my question, I guess, is... it's not a question, or comment and question, I think, is where do we determine sort of the levels or the definitions of AI use, human use, those sorts of things, given in filmmaking, as you just pointed out, you know, there are so many different layers. How do we actually, you know, navigate all of this as makers and as producers? It's still very nebulous, let's say.

Angela Dunning 00:53:11

Thank you. Do you want to crack at that, or should I? I'll take a quick crack. So I regularly work with directors, directors of photography, big studios, independent creators, and the tools are great for certain things, and they're not great at all for other things, right? I have yet to meet too many directors of photography who would turn over to AI the creation of certain scenes, It, it, it, I don't see that happening anytime soon. For in the US, the key to being able to own the output of what you create, even when AI is used, is human control. Are you actually exerting the creativity that is dictating what the final piece or image or text will look like? And where you're using AI to enable your human creation, that is going to be protected, and that should not change the copyrightability, your ability to own that. Where you are giving over to AI and simply entering simple prompts and letting it do the creative work, then whether we're talking about copyright or patent or anything else, that's not going to be deemed human authorship or protectable.

Erik Svilich 00:54:26

Yeah, we work with a lot of, or talk with a lot of regulators, standards bodies, and just industry associations. All three of those groups are asking that exact same question, and nobody's figured it out. It's very much a grayscale as well, kind of like what Angela was saying, is it's a tool, and depending on how much control you're giving over to the tool, you know, that may matter more to some and not at all to others, right? So it's also like, does the audience care? You know, and so it's... nobody's figured it out yet, and it's an active conversation for us all to figure out.

Carlo Rizzo 00:54:57

Yeah. Figure out.

Audience member 00:55:01

Thank you all, because I think it's been a fantastic discussion. And Phil from Goldfinch and also Engine that Carlo kindly referenced and is featured in this fantastic report, but more of a comment than a sort of straightforward question maybe, but... and Elisa kind of touched on this right at the end, which I thought was great, which is the technology in this disruption that we're seeing ultimately needs to be used for the fans' benefit, for the audience's benefit, because we can... and for sure there's an opportunity and then there's a huge need at the moment for infrastructure to be built around all of the fantastic comments that have been made. But we saw this in Web3. Everyone piled into Web3. It's all about community engagement, engagement, creator tools. And actually we came at it from a different angle with a platform called Myco. We grew that in 3 years from 1 to 40 million with the idea of rewarding the user, rewarding the... putting the power back in the hands of the audience. And I think there's not a worry about it, but there's definitely, you know, we definitely need to think with the audience in mind, I think, across all of the industry, but definitely when it comes to the tech side too. And that's where I suppose Enjin for us comes from, creating the rails for fan fiction and participation. Yeah. Anticipate VIP to happen. Because if we're not thinking fan first, then, you know, kind of why are we doing this at the end of the day? So again, more of a comment than a question, but thank you, Elisa, for kind of at least kind of starting the ball rolling a bit as well.

Carlo Rizzo 00:56:32

There is one more question over there, and then I think probably if no one kicks us out, we'll do another one here.

Yannick Bossenmeyer 00:56:45

Who is the... oh, sorry.

Angela Dunning 00:56:50

Do you think current AI and IP frameworks risk benefiting large platforms more than independent creators since large companies can afford legal uncertainty and licensing negotiations?

Carlo Rizzo 00:57:04

I guess, I don't know, Yannick or Angela, you want to... Can you take that?

Yannick Bossenmeyer 00:57:11

Yes, of course. The big companies have more money to pay lawyers, good lawyers like Angela. But that's why we have regulators and why public governments and so on are working on the subject. So I'm confident that there will be solutions.

Angela Dunning 00:57:28

In the US, we have about 100 lawsuits pending. Copyright class actions, there are trademark claims, there are breach of contract claims around scraping, the whole ecosystem is covered. Not one user of AI tools has been sued for copyright infringement, right? Because at the end of the day, every one of us remains responsible for what we make and what we put out into the world. And as long as we continue to demonstrate and exert the care and attention to the rights of others that we should all have been applying all of this time, I think the chances that that uncertainty comes to individual users of the tools is much lower.

Audience member 00:57:28

Hello everyone. Thank you for the wonderful talk on AI. It's very interesting to see how it's changing the film landscape. So I wanted to ask you, I would like to ask, how do filmmakers usually credit AI? And should film, like, what is the scale of when AI should be credited and when it should not be credited?

Erik Svilich 00:58:46

Well, good question. It's currently mandated by law in the EU by the EU AI Act coming into effect this year, even if there was just a small modification or edit. To your film, that needs to be machine-readable metadata embedded into the file itself. It cannot just be part of the credits. So there are open-source standards, like C2PA, the organization that I work with, to enable creators. And there's some platforms, like Adobe's tools, that do that automatically. Or there are third-party vendors and APIs that enable you to do that as well. But it's a requirement by law nowadays. Thank you.

Carlo Rizzo 00:59:24

One last question here in the front, and then thank you all for your patience as we run over a little bit of time.

Audience member 00:59:30

Thank you very much indeed. We're doing a film that is going to be maybe about one-third AI, pure AI, a lot of it just regular filmmaking. And one of the things that people keep on telling me is that if we use a Chinese company or one of several Chinese companies, that a distributor, like someone who we're going to sell the movie to in the end, might have serious issues with that. And obviously there's different layers. You might have your own servers and keep everything on it. And the other small wrinkle on this is that it's 3rd century Rome. So it's not like we own the copyright on the Colosseum, or... you know what I mean?

Audience member 01:00:10

Yeah.

Audience member 01:00:11

So how does one get around that? Is like Netflix and Amazon going to go, Sorry, you know, you used a Chinese thing, so we can't buy it.

Angela Dunning 01:00:22

So with the proviso that I'm not dispensing legal advice from the stage, I would... the first piece of advice would be to consult a good copyright lawyer. But not being glib, I don't know that the tool you use is the critical distinction. As always, it comes down to whether you can demonstrate the human-authored content in the work, right? If there may be some part of the production that is AI-assisted or AI-generated, but where the script or the overall work is human-authored, then that should not be a barrier to being able to license and monetize the work.

Carlo Rizzo 01:01:09

Thank you, Angela. Thank you, Janick. Thank you, Erik. Thank you, Elisa. Thank you, everybody, for staying with us, and enjoy the rest of your day.