Verge 访谈 RIAA CEO：从 Napster 诉讼到 AI 音乐版权与标注之争

Today, I’m talking with Mitch Glazier, who is CEO and chairman of the RIAA.

This conversation was a real full-circle moment for me: I was in college when Napster hit, and the RIAA became everyone’s villain for suing kids for downloading music. That was when I decided to go be a copyright lawyer, which set the rest of my career in motion — and Mitch was in the background as a young lawyer at the RIAA working on those cases. So part of me was very deeply, very genuinely surprised to discover just how much fun I had talking to Mitch.

That’s because here in 2026, Mitch, the RIAA, the record labels, the artists, and ordinary people are all pretty unhappy with the big AI companies in the same way, for training on everyone’s work without permission and flooding our feeds with slop.

Verge subscribers, don’t forget you get exclusive access to ad-free Decoder wherever you get your podcasts. Head here. Not a subscriber? You can sign up here.

Everyone is mad about it! And since the RIAA is both the group that lobbies for changes to the law and also the group that files lawsuits about things that affect record labels and distributors, Mitch is a pretty busy guy these days. We got deep into the legal complexity of things like AI deepfakes, how banning them might affect other kinds of expression, and how the AI detection and labeling systems that we all seem to want might actually work in practice — especially if big platforms like Spotify and YouTube don’t get on board.

I just want to get right into it. This conversation is a real ride through how AI, the law, art, and commerce might all intersect. I think you’re going to be able to tell, Mitch and I had a lot of fun talking about it.

Before we get going, do me a favor and subscribe to Decoder on YouTube, where we put out new episodes every Monday and Thursday.

Okay: RIAA CEO Mitch Glazier. Here we go.

This interview has been lightly edited for length and clarity.

Mitch Glazier, you’re the chairman and CEO of the Recording Industry Association of America, which people know as the RIAA. Welcome to Decoder.

Thank you. Thank you, Nilay. I’m excited to be here.

I am very excited to talk to you. I have to tell you, this is a very full-circle moment for me. My entire career got on track when I was in college. Napster came out, my university banned Napster from the school network, and I decided right then I was going to go be a copyright lawyer. And then I was a copyright lawyer, and I defended kids who got sued for using Kazaa. I was not good at that job. It was not fulfilling. And I became a journalist because I could write about copyright law on the iPod. So talking to you is about as full circle as it gets for me. Thank you for being on the show.

Good. Well then, I’m no longer sorry for the policies that we forced your school to implement because it made you a copyright lawyer. So the world benefited at the end of the day. It had some silver linings to it.

When I first started at the RIAA, Napster was obviously the big issue. Nobody had seen anything like it before. The last technological change had really been CDs. I mean, it had been a long time, a much more gradual change and much less abrupt. So they decided it’d be a good idea, since I was youngish at the time, to make me go to college campuses, go to auditoriums, and convince these college students that they shouldn’t use peer-to-peer technology.

Wait, wait, wait. I imagine when you said that, it was more like a 21 Jump Street situation, where you were posing in the dining hall being like, “We should pay for it.” Were you giving presentations?

Yes, it was horrible. Everybody had these anti-RIAA t-shirts on. I remember one time I followed Chuck D, who was there promoting peer-to-peer. And I didn’t know what to do, and I knew I was going to get booed. So I took a garbage bag, put it over my head, and I walked out onto the stage, thinking they were going to throw tomatoes at me; I might as well make light of it.

But I have to say, even though it was a very tough time, some of the conversations that we had with students at that time were some of the best conversations I’ve ever had — even now with AI — about the change in technologies, things being fan-driven, and the disconnect that can occur sometimes between an industry and the people who are out there driving change in the industry. So it was actually a good experience in the end, but it certainly did not feel like it at the time.

It didn’t feel like it at the time for me. Thank you for saying the butterfly effect is real. But it was heated. It was hot.

One of the things that has really occurred to me as I get older, and AI exists, and I see the platforms taking so much of the economic value out of art, is that my opinions about intellectual property enforcement have changed. It feels like I’m just giving up all of the ideals that I had when I was a kid about the internet, that open access to information was good, that… There was that meme ages ago: everything is a remix, and we should all build on each other’s art as much as we want because that will push the culture forward. I am still very idealistic about that. And I definitely did not think the recording industry suing kids was the appropriate way to protect the intellectual property of artists back then.

But here we are in the age of AI, and it feels like no one is making enough money for making art except the world’s biggest artists — who are all also tech investors now in some way — and the tech platforms that collect fractions of pennies from everything and are doing data center build-outs or whatever it is. And that just existential dynamic for me is where I wanted to start.

I want to talk about how the RIAA works and all the Decoder questions, but I was just thinking about what I really wanted to ask you, and it comes down to this: we have to protect the work, but there’s some balance in overprotecting the work that makes it harder for people to make new things. And my own journey with the RIAA is a reconsideration of that as more and more things get taken.

Where have you landed? You’ve been there for more than 20 years. How are you feeling about that balance?

Yeah, so the word balance is the right one, and it’s not an easy one. It’s one of those things where, over time, a barrier pops up here and then you kind of pivot, and then a barrier pops up here and you kind of pivot, and you sort of land in a zone that feels right because the responses that you’re getting, both from the creative community and from fans, start to line up. And I think in AI, they are lining up much more closely than they did during previous years of disruption.

Some of it has been because of the democratization of the industry, where the gatekeeping roles have changed, where the partnerships between those who partner and invest with artists and the artists themselves have changed. In a world where artists and label services, like à la carte services, are a bigger part of revenue, in an area where catalog, forever exploitation, and new investors in the industry have come in, the landscape is much wider now. And it’s not like there are just two types of big actors controlling or sharing in all of the money.

Now there are dozens of different types of folks meeting artists in different places where they are and meeting fans in different places where they are. And so it feels to me like this disruption, which is huge, and I hope we talk about it… The good news is, unlike Napster, I don’t feel like we are facing it at odds, which is kind of what it felt like back then. Now it feels like we’re not in the same place, but we’re kind of rowing in the same general direction, which is that everybody understands that art and commerce have always had tension, and there’s always going to have to be a balance. But authenticity and that connection between a human being artist and a fan, the preservation of that seems to be paramount in everything that we’re doing, whether you’re investing, listening, or performing.

And to me, we’re not fighting in this space anymore between just a business model or an investment. We are now fighting to keep the art as human and the connection between an artist and a fan preserved and alive. And that to me just seems like a much larger purpose, I guess. So it feels very different to me. It feels like a good bookend.

Like you, I started my career with Napster, but I was on the other side, learned a ton as a young lawyer, and now on the other side, we have AI, and comparing the two is so wild. I mean, the polls during Napster were like fans over here, and creators kind of hiding a little bit, and the companies over here. And now it’s like AI’s over here, and we’re all together on the other side.

Yeah. I mean, again, I cannot emphasize this enough: I was a bad lawyer, and I should not have been doing that job.

[Laughs] I doubt that.

I was not good at it. I really wasn’t. And what I remember from that time was that it was college kids; their universities would turn over their IP addresses, and they would get their little lawsuit. We would tell them, “We’re going to go to the district court in Chicago; you’re going to settle for $5,000, because that’s what the RIAA is running this program at. It’s a $5,000 settlement to break even on the cost and to deter you, and then you’re going to go work at the pizza place for a semester, pay your bill, and move on.” And I thought, this can’t be accomplishing a goal. The thing I’m doing right now is nothing. And maybe that’s why I felt like I was a horrible lawyer. Maybe that’s why I got out of it. But what you’re describing is the poles there, where there was no pushback on, well, these kids just love music. The thing they love is just music, and now they can get it, and they have it. They weren’t thinking about the business model of CDs.

That’s what Napster fundamentally disrupted. There was a scarce physical object that you could go pay $15 for, and if Metallica sold enough of those, their plane would get bigger, and everyone could kind of see that. And, “I’m listening to a lot of this music, and it should make you rich,” was nowhere close to reality. We’ve built those models now. We have built Spotify and Apple Music and all the subscription models. I want to talk about YouTube quite a bit, actually, where you can listen to a lot of music and some people will get rich, but a middle class of artists may just be totally hollowed out. And now the AI companies are like, “And we’re going to set up a competitor where you can just ask for music and no one except us will get paid.” And that is probably built on a bunch of artists that used to make money by having a lot of attention.

That swirl for me is the one that I can’t decide where my ideals fit into. As a person who has opened Suno and had a good time just making some music, I’m like, well, I understand how this is great for one kind of person with an artistic bent. I understand why this is fun, as somebody who likes music. I do not understand the economics that follow from this, the same way I had no idea how the economics of disrupting the CD would shake out. I’ll actually give you one example just to make this question even longer. We had the CEO of Cloudflare on the show, and they’re trying to set up a system where, instead of websites running advertising on traffic, everyone gets paid a fraction of a cent to make content on the web, like Spotify.

I was like, “Dude, Spotify changed the entire music industry. Artists have different jobs now because of that economic model.” And he was like, “No, I think the big ones are really happy.” And you can argue with him or not argue with him, but it felt to me like the tech industry did not understand what that disruption actually did to music at that moment. Now we’re standing on the precipice of it again. So again, I just want to start there. I’m feeling this existential tension; you described it as art and commerce, where commerce is going to change, and so art has to change as well. How are you thinking about that?

Since we’re talking about existential things, if you go back to why copyright exists, it’s to create more opportunities for more art in the future. That’s why copyright exists. That’s the reason why the property right exists. It’s to create an incentive to allow people to protect their creations so that they will go ahead and make a living doing it. We don’t want this just to be a hobby. We want this to be… Artists are extraordinary people in our culture. They are geniuses who need to be incentivized to share that genius because they’re not normal people. They’re not me. These are incredible people who have something to share, and if they can’t make a living doing it, then they can’t and won’t share it, and we are all worse off. That’s the whole purpose of copyright, is to be able to allow them to be able to benefit from that in some way, shape, or form.

Now, obviously all this infrastructure has grown up around it, all these companies have grown up around it, et cetera, but that’s sort of the basic premise. And you have to ask yourself the question: if non-human music, which can be made in two seconds without lived experience and, in my view, without a soul, can dilute or drown out the incentives for people to be able to make a living as humans making art, what happens? What happens to our culture? And I think that is an existential question that is not just academic to ask.

We could talk about differentiation in the marketplace; we can talk about whether or not Taylor Swift competing with an indie grunge band is the right model for everybody to be able to get their piece of the business on these streaming services. But at its core, it’s still essentially human, and it still distributes art that took sometimes 20 years to make, sometimes three minutes to make. It depends on the genius of the artist doing it, but it still has all of those human warts and craft. It has craft.

I think that if you have a system where the incentive is with the tech company that makes the AI to use everything that humanity has ever created to make money and, at the same time, then deplete the opportunity for future human music, we’re going to have a society that is one way going backwards and another way going forwards. If the culture suffers going forwards more than improving on what went backwards, we are in the wrong place, and we’re not doing the right things. And that’s what I mean about sort of rowing together.

I live in the policy world. You have to ask yourself, what policies can we put in place to create the incentives to allow that better future of human-made music? It’s really hard. It’s super hard.

When I say I’m having this existential dilemma… Just to put it into clear focus for the listener: In the age of the CD, the economic value was tied to a scarce physical object. There are only so many CDs. And then you could maybe copy CDs, and Apple would get in trouble for saying that you could rip, mix, and burn CDs. That was a whole ad campaign, and Apple got in trouble for incentivizing people to copy CDs. But you would pay money for an object that existed, and then we got to the digital world, and the friction to copy the song went to zero. We had to invent a new model, and that model looks like Spotify and the rest. Now we’re in the age of AI, and the friction to create might be going to zero. So there’s even less scarcity because you can just horsepower an AI pop star into existence singing an AI song, and there’s going to be a glut of supply.

Somewhere in there is marketing, somewhere in there is fandom. There’s all this stuff that can create other kinds of scarcity, but that’s the journey the music industry has been on. I always say whatever happens to the music industry happens to everybody else five years later. The music industry is always on the cutting edge.

Where does the new scarcity come from in this model? Is it you doing policy work and lobbying governments for changes to the laws? Is it fans revolting and saying, “We want human artists”? Is it the platforms labeling everything? Where does the scarcity come back in?

You’re the first person, I think, that I’ve listened to who has honed in on this particular question. At the end of the day, it’s always the fans, in my view. They will always win, because if you are any kind of service, with any kind of business model — I don’t care if you’re AI, I don’t care if you’re streaming — you need human beings to pay you money, usually by subscribing, as that’s the highest revenue source. And they’re not going to do that if they feel like they are losing out. I do think, given all of the focus groups, polling, everything that we’ve done, and other people have done on this, it is very clear that fans want to maintain a human connection with an artist.

It’s not just about the music. It’s also about the artist, the persona, the brand, the emotional connection. Taylor Swift’s fans love her music, but they also love her. Pearl Jam’s fans love their music, but it’s also, like, Pearl Jam. You know what I mean? They’re their own piece of culture. And so we know, for example, that if a fan knows that a song is AI, then they have a different view of it than if they know that it is human.

Do you think that’s going to last forever? I’m honestly curious about this. It has changed even in the last year.

It has. I think that, with generative AI, it will last forever. In other words, with a completely machine-made piece of art, it will, because you lose that connection. On assistive AI, where creators are using AI, even in a broad way as a tool to get their creation out there, right now it might be a little bit suspect. In a few years, it may just be like using a synthesizer or using Auto-Tune. There might be some people who think it’s not the greatest, but it’ll be an accepted part of the art form. So I think there’s a difference between… Are you maintaining the human connection? That’s the question. You can use AI as a big part of the tool to do that, but in some way, are you maintaining that human connection?

I think that’s the test, and that doesn’t happen with a prompt. I mean, that doesn’t happen with me sitting there typing in 18 words on Pseudo, then more words to get it to a place that sounds better to my ear, and then sticking it up on Spotify through DistroKid. I just don’t think that will ever cut it, even though I think there is a bigger scale when it comes to assistive AI.

So, fans rule. I think labeling is happening and will continue to happen because of fans; I think consumer reactions will happen because of labeling, and that will then drive the economics of the business. So labeling is actually a very important piece of this. But I also think that if you don’t secure the rights, fans never get a chance to then do their thing, which is to ultimately control the market. My focus all around the world is: have we secured the basic rights of creators? If we’ve done that, fans can take over and drive the market from there.

I want to talk about all that. It occurs to me that I have gotten immediately as deep into my feelings as I can with you, and I forgot to ask you the very basic questions that I ask everybody who comes on the show. So obviously the RIAA is well known; it’s either a hero or a villain, depending on where you are in your story, as we have both determined.

Tell people what the RIAA is, how big it is, and how it’s structured.

We’re based in Washington, D.C., because that’s where the federal government is. We used to be based in New York. RIAA actually started as a tech standards association. They created the RIAA curve, the equalization curve that allowed both the recording and the playback of vinyl records in a way that would work on every single record player. So it didn’t really start out as a lobbying organization or a litigation organization; that’s how we became famous or infamous, depending on where you sit. We started out as a tech standards organization, and we still do that. We still participate in international standard-setting organizations right now to try to set the fields, for example, for the metadata that people will use, whether it’s AI or any other technological format, or in labeling, for example. But we also run the gold and platinum program, which we’re probably best known for with the general public.

So we’re like those nerdy auditors that actually certify the gold and platinum records, like the PricewaterhouseCoopers lawyers that walk out on the Oscars or whatever. But that’s great because it allows us to interact directly with artists and celebrate their success, which is very cool. We also put together all the revenue from the entire recorded part of the music business, and we release that twice a year. That helps inform people about the markets, and it helps to, on the business side, give people the benchmarks that they need industry-wide. Everybody knows what their company is doing, what their management company is doing, but to know what the whole industry is doing and whether or not you are on par is quite helpful.

We are the spokesperson when the industry is aligned. We’re out there talking on behalf of the industry. We technically represent the major labels, but we forge coalitions because whether you’re major or indie, whether you’re an artist or a publisher or a songwriter, most of the time our interests are aligned. Every once in a while we fight with each other, but when we are aligned, which I would say is 90 percent of the time, the RIAA is sort of the driving force in building coalitions so that we can all speak together.

So we do a lot of stuff. We don’t just sue people and lobby the government, although those are big parts of what we do. And then we also take on the content protection force for the industry. Our content protection department operates 365 days per year, 24 hours per day. We work with sister organizations in London and New Zealand. So it doesn’t matter what time of day it is; we have people coordinated and working around the clock. And then we do all those take-down notices. We get a note from a manager saying, “A pre-release of this artist’s material that hasn’t come out yet is up. We need your help getting it down.”

We do all of that, and we do it… The entire organization is only 46 people.

Wow.

Isn’t that amazing? But the most passionate music lovers you would ever imagine.

It’s only 46 people?

Only 46 people.

How is that structured? Does everyone just do everything all day long?

There aren’t as many redundancies as there should be. We’re divided into those different departments, like content protection, communications and research, public policy and industry relations, and obviously legal affairs. So we are, I would say, small and nimble, but we punch above our weight. We are high profile, but our industry, if you think about it compared to the folks that we are up against, if you want to view it that way, is really small. I mean, the entire recorded music business was $11 billion wholesale last year in the United States. The market cap of just one AI company is $2 trillion.

So when we are out there lobbying, and we’ve got our three-person lobbying team at the RIAA trying to make a point to policymakers, and you’re dealing with super PAC money that equals $250 million in a midterm cycle, we don’t have that kind of revenue. But what we do have is music, and we have artists who everybody loves, who are culture makers, and who are messengers that people trust and listen to. I think we do pretty well for our tininess.

The other question I ask every CEO who comes on the show is about decisions. You have a lot of decisions to make. Some of these are decisions of first impression. No one has ever decided how AI and art should work together. How do you make decisions? What’s your framework?

Personally, my leadership style is that I like to have people around me who are not “yes” people. I don’t want people just sort of… If I come up with an idea, I don’t want it to be reinforced immediately. If you agree with that, great. But I like to hear from everybody. I think my team knows that I like to hear the truth, and I want to hear their raw opinion because I want to check my instincts. I usually have a gut instinct, but then I will survey people that I trust in our companies and people on my own team. I’m learning to be a better listener. When I hear people, as soon as in my mind, I feel like I know what they’re saying, I tend to then jump ahead to the next… I’m deposing them. I have a million questions.

I’m learning to be a better listener and kind of understand where they’re coming from. Then I take all of it, and I synthesize it. And I would say about maybe 60 percent of the time I end up with my original gut reaction, but 40 percent of the time I take a different path because of what I’ve heard and the expertise of those who are around me. So it takes a little longer than it maybe should, and you don’t always have time. Sometimes you have to make a snap decision, but on something really big, like what strategy are we going to take? If we feel like we need to bring a lawsuit, which is going to be a major, publicized, precedent-setting lawsuit that could determine the rights that creators have, you have to take that as seriously as it is, and make sure that you are getting all the opinions that you need to get.

I also have a ton of friends who work, so to speak, on the other side. These are people who I trust, and I’ve known for many years, and I will often, without disclosing anything I’m not supposed to disclose, call them and say, “What is your perspective on this? I live in this world of creator rights where I’m thinking in one direction, but what would you argue on the other side? How are you looking at this?” And when that informs your decision-making, not only are you more prepared, but sometimes you will alter the way you think about something by putting yourself in their shoes and understanding there’s another side to this that’s going to make sense to a judge or to a policymaker because they’re not crazy. They have a different take, and it’s pretty valid.

Sometimes I have to be a moderating force even in my own industry, saying, “We’re living in our bubble, they’re living in their bubble, and so if we’re going to be effective at finding a solution, if that’s what we want, then we may have to think about this in a way that leads to a landing zone.” If we don’t want a solution and what we’re trying to do is make good trouble, that’s okay. There’s nothing wrong with that. That’s a good thing. It can lead to good results. But if we’re trying to get to here, we better figure out a realistic path to get there.

Okay. I’m very sorry for this, but it feels to me like the rest of this conversation is just going to be nightmare law school hypotheticals. This is just what we’re doing now.

Nothing wrong with that.

That framework applies to a lot of different problems. Particularly on the other side, they make some good points too. So I opened Instagram today, and I saw nine million deepfakes of celebrities being put in the video for Pop Star by Drake. Have you ever seen this video?

Yes, yes.

Drake didn’t do it himself. It’s Justin Bieber acting as Drake in the video. And I saw one with Lewis Hamilton today. I saw one with Donald Trump today. I saw a full Star Wars one where it’s Darth Vader, and he’s just been deepfaked into Justin Bieber’s mansion singing a Drake song. And I thought to myself, I can’t even count the layers of intellectual property issues here. First of all, no one has the license to any of this likeness, right? Lewis Hamilton could probably be like, “Hey, that’s me.” I don’t know under which law he gets to show up and say, “Hey, that’s me,” and to whom he says it. Is it an AI company? Is it the person who made the video? Is it Instagram? Some of the platforms have private policies like YouTube now. Even just that, where to begin? That’s very connected to you saying the scarcity is in the artist.

The scarcity is in the relationship the fan has to the artist, but that has nothing to do with the copyrights you routinely enforce, which is the recording. You can see I’m just spiraling. I watched one video, eight seconds of Darth Vader doing a Drake song, and I thought, I don’t even know where to begin. Is it Disney who gets to be the angriest, or is it Drake, or is it someone in the middle? Can you help me just organize my thoughts here?

First of all, I’m so glad you’re bringing this up because we literally, over the past three years, have been dealing with exactly this. Because believe it or not, and it’s so bizarre to think about, none of us right now has a right to our own voice or likeness under federal law. Most of us don’t even have it under state law. State law, which is called the right of publicity, usually only applies to advertising or endorsements. I couldn’t take your face and put it on a Wheaties box without asking you because I’m selling it and making money, and I’d be using your popularity to do so. Those are very limited rights. Now we’re in the world of deep fakes where everybody can do anything, make you say or do something you never said, you never did, and that you might not approve of — and we don’t have a basic right to ourselves under the law.

It seems so weird in a country that was founded on individual liberty and individual rights that we wouldn’t have an individual right to ourselves. So we had to sort of start from scratch. And what we did was get together with YouTube in the very beginning. These things, even if they’re not created by the platforms, the avenue to the fan is through the platform. And everybody’s not going to bring a lawsuit against everybody in order to get this stuff down when it’s proliferating as fast as it is. So the question was, what responsibility does or can the platform have to give a very simple takedown mechanism that anybody could use? Because this is your face, it’s your right. It’s not like the record company owns the right to your face, or the movie company. You own the right to your face.

So if you don’t like the fact that somebody used your face or your voice in a way you didn’t like, you should be able to very quickly say, “That’s me, I can prove it’s me, and I don’t want it up there.” And the platform should want to do something to protect the integrity of the platform without liability.

Can I just stop you? This is why it’s a nightmare hypothetical, right? A longstanding trope at The Verge started by our former colleague, Sarah Jeong, is that copyright law is the only functional law on the internet. The only way you actually really want to regulate the internet is to take things down. That is the thing that people want to accomplish. The First Amendment basically prohibits the government from creating some mechanism to take things down, except in the case of copyright law, where we have the Digital Millennium Copyright Act. There’s a huge notice-and-takedown regime in that law, and it is frequently abused on all of these platforms to get things taken down.

The purpose of that law has been stretched beyond recognition. If you issue a YouTube strike to a YouTuber on that platform, all hell breaks loose. That is an act of war because unless you’re doing it for the right purpose, they know you’re doing it to harass them.

This is a lot, right? You’re describing a new notice-and-takedown regime that isn’t based in the law. It’s based on some private company’s understanding of what a likeness is. And there are all kinds of reasons that I might want to publish a video of Donald Trump. There are all kinds of reasons that I might want to publish a video of Drake that have nothing to do with using his likeness for commercial purposes. And granting those people the ability to take things down is the same problem without, I don’t know, a federal fair use regime or some other… How do you balance that?

Great. So this is why it took three years, and it’s still going on, because what you’re saying is, first of all, you need to learn from the experience with the DMCA. It is the precedent we have, and there are obviously problems on both sides. You need to learn from those. The second is you need to create a new intellectual property right, in this case, much closer to trademark than to copyright, but still an intellectual property right, because that is, as you said, sort of the functional way of balancing that against the First Amendment. If you have an intellectual property right, you are allowed to remove it despite somebody’s other free speech rights. In this case, there’s a weird analysis of, do you have a free speech right to control how your voice is used? So it’s like free speech versus free speech.

What we did was, besides getting together with many First Amendment law scholars and trying to hammer this out, which was a fascinating learning process, by the way… I feel like I got to go back to law school and listen to these brilliant constitutional scholars talk about the tension between the First Amendment and intellectual property rights. What you have to do, if you’re going to create a takedown regime, is put in hurdles on both sides. You have to put in hurdles on the “I’m telling you to take it down” side. And there are lots of things we’ve learned from the past, like: Are we going to force a, in this case, First Amendment review before you can send the notice? What’s the punishment if you send a false notice? Is that punishment high enough to deter false notices? Then on the back end, you have to have a counter notice, right? And then how do you make sure that that works well and that it’s not too low a hurdle, but a little bit of a hurdle because you want to have balance on both sides?

And then finally you have this middle area where there’s going to be a decision about whether or not it’s a parody, for example, which many are, and the First Amendment trumps copyright when it comes to a parody. What if it’s news? And what is news anymore? What is bona fide news versus some of the stuff I’m getting when I’m scrolling? We then created this middle period where a temporary restraining order, a very quick legal action, could take place and a judge could actually decide the constitutional question — because nobody wants the platforms deciding constitutional questions. The people on both sides are like regular people. They’re not going to be making ultimate determinations. So that’s ultimately what we came down to in the NO FAKES Act, which is the name of the federal bill that’s now being pushed. And whether the balance there is perfect or not yet, I don’t know, but I do know that we learned a lot of lessons from the DMCA.

I think everybody is very aware that the First Amendment defenses are really paramount in any kind of regime like this, and that overprotecting on either side… If you overprotect on the First Amendment side, then this also applies to a girl who’s bullied in high school and some kid takes her face and sticks it on a body that’s not hers. You don’t want that staying up when it shouldn’t. On the other hand, I have the right to use your face for certain things because that’s my free speech, and it’s really funny, or I’m making a point about underlying character or something like that. So, long answer: there’s no perfect response, but this time we had OpenAI at the table, we had YouTube at the table, we had people from Public Knowledge at the table who bring a different perspective than we would bring. We ultimately got to a place where I don’t think folks are thrilled on all sides of it, but I think they think it’s progress.

It’s something. I will grant you that it’s something. It’s better than what we have now, but it’s fundamentally the creation of a new intellectual property right.

Absolutely it is.

And there’s some balance there that almost certainly will track the same balance that copyright law and music have undergone over the past hundred years.

No question.

The initial debates about copyright law and music were about player pianos. We’ve built huge institutions to manage these rights.

I said I was going to talk about Google a lot, and the tension inside of Google when it comes to this is the next sort of nightmare law school hypothetical for you. Google has two conceptions of itself. YouTube describes itself as a licensing regime. There’s a lot of content on YouTube; the creators make it, and Google takes a license to it. Whether or not any creator ever gets to negotiate that license is of secondary concern. There’s a lot of music on YouTube, and the industry had to fight its way into licensing on YouTube. Music is existential for YouTube, and the industry knows it has that power. There’s a real back and forth. Maybe the only industry that has true negotiating power with YouTube is the music industry. And you can see this play out over and over and over again.

On the other side of the house, Google Search is a huge business; they will describe it to you as a fair use regime. We’re going to ingest all the photos in the world and let you search them; that’s fair use, and we’re going to litigate our way through that. We’re going to ingest all the books in the world for Google Books; that’s fair use, and we’re going to litigate that. And they have done that over and over and over again.

You can work with the people at YouTube. They’re very reasonable. Neal Mohan is a smart guy. He understands that he needs music on that platform to survive, and he will build you the tools that you’re asking for, and he has. And then the people running Google are like, “We’re just going to eat everything, and training an AI model is fair use.” And just in that one company, they are totally at odds with each other. They’re going to kill each other somehow. How do you make sense of that?

What’s so fascinating to me about YouTube is that it is the first real lesson in the arc of the DMCA. It’s where I think a technology-based company realized that even if you win the fight on intellectual property rights, if you don’t have a partnership through a license in the actual market where fans are, you are not going to have the service that fans really want. And so it was the first time that everybody realized that you can win the case but lose the business. Music is such a driver of YouTube, and I think this applies to AI too. Having pristine copies that fans really want, and then having marketing partnerships and partnerships with people who invest in and partner with creators, is what matters. Those creators have a better sense of how to create the bond between that particular artist that they’ve been living with and sort of building a brand around. They understand that, versus sitting in a cubicle in a tech company and trying to decide how to connect the fan with the artist in the best way, that the former piece is irreplaceable.

And I think YouTube realized that and developed Content ID even though they technically won in court and didn’t have to, so to speak. Now, the partnership is so ingrained that there’s a dependency, or a codependency, and that’s how it’s supposed to work. It also works because of this concept in copyright law called substitution. If what you’re creating substitutes for something else, it’s probably not fair use. That’s what we’re arguing in our case against Suno. You’re basically creating music that is competing on the same platform right next to human-made music; you can click on either one, and you’re paying the same subscription fee. And so you’re affecting the economic pool, and you’re effectively substituting in the marketplace.

This is the fourth factor of the fair use regime.

This is the fourth factor of fair use, which courts say is the most important: are you affecting the potential market for the underlying work? And if you are, that’s sort of problem number one. Google, unlike in the YouTube case under the DMCA, the case that they won in the Supreme Court was about creating an index. And the idea is if you click on it, it could lead you to the work that you might then buy. The search result doesn’t substitute for the work, but it could lead you to the work, which at the end of the day may not have the same economic consequence. It could even have a positive economic consequence. At least they would argue that.

So there’s this weird thing: Google won its case, and YouTube won its case, but YouTube formed a partnership despite the win because that’s the best fan experience. And Google didn’t because it wasn’t as related to the fan experience. The fan experience was potentially pointed to by their activity. So that’s my weird law school take on the two cases.

If you’re running Google, if you’re in charge of Google Gemini, you need all of the data in the world. That is just how it works. To make the models better and to win against OpenAI, you need more data, and all of those companies just took it. I do not think Google came to you and said, “Can we license all the music in your database even though we already have a license on YouTube?” In fact, they just scraped YouTube.

I don’t think creators have woken up to the fact that they’ve given a lot of economic value to Google and its models. The other AI companies also scraped YouTube, and Google can’t yell at them because then they will have to reveal that they scraped YouTube, and this is about as tense a fight in the AI industry as there is.

You’re going to come to Google at some point, and you’re going to say, “No, you got to pay us for that training.” I know you’re suing Suno, but that feels like a little test balloon for the people with the real money. At some point they’re going to say, “No, this training was fair use.” We’re watching this dynamic play out in dozens of these cases across the industry. How do you make sense of that? Where you’re going to say, “You’re going to kill us. If the AI models flood YouTube with fake music, you’re actually going to kill the supply of the data that you need.”

So there’s the legal side and the business side. On the legal side, obviously we’re bringing this as a precedent-setting case to show that it is not fair use to train on music. And whatever exemption you want to think of, it’s hard for us to imagine how it applies. I don’t think national security is at risk because you have to copy Drake. I don’t think that there’s-

Wait, what did we play in the war in Panama? We played Metallica at them at full volume?

Right, right. Maybe it’ll be used for soft power, or propaganda blasted on the screens.

The Clash was very upset when “Rock the Casbah” was the soundtrack to the Gulf War. I’m just letting you know.

I bet. I bet. Going back to securing the rights. We’re bringing this case to make sure that the rights are secured, so then every company knows that you are not allowed to make reproductions of other people’s music without consent or compensation. That’s basically what we’re trying to establish. And then once you secure the right, you let the marketplace develop from there. I’m sure that the good people at Google and at other companies understand the risks. It’s fascinating.

Nilay, I don’t know if you’ve read the SEC document that Anthropic just had to file in order to have their IPO, but in the risk section, it’s basically like, “Well, it could be that all this scraping that we’ve been doing actually violated all these laws and that could, if it’s willful, put us in a liability deficit of x. People who are investing as new shareholders should sort of know that information.”

So they kind of acknowledge upfront this could be a high cost of doing business going forward, and yet they’re going to have a spectacular IPO despite those risks. This means that they’re willing to bear them, and they’re also willing to accept the cost if and when that comes and it is secured.

Actually, I have a different view on that, and it’s related to the beginning of our conversation. I think you gave a quote to the Los Angeles Times where you said being the RIAA is more fun now because the consumers are on your side fighting the AI giants instead of you fighting consumers. I get that. The valence of the RIAA and the culture may have changed. I’m sure we’ll get some people in our comments saying it hasn’t, but in general, it’s changed, right? You’re on the side of the artist fighting the giants.

Google won all of those cases about fair use when it was a baby, right? And they had slides in the office, and they were a startup, and no judge had ever used the internet before, and they’re like, “I love YouTube!” And Viacom, you’re posting your videos on YouTube because your people love YouTube, and you know the audience is there. There was something very happy about the tech industry at that time. Google, as a company, was a hero.

I think these companies are villains now. But truly, and if you look at the polling, they are not beloved. AI is not beloved, particularly with young people. The idea that AI is theft is held by young people. That gives you an opportunity to actually turn the screws harder, to say maybe you shouldn’t bear these risks. Maybe the artists should be the ones getting rich, not the AI companies. Are you putting that into policy, or are you just trying to negotiate to an outcome?

No, we’re putting it into policy. Everything is an arc, as you know, but in the beginning of a disruption, you try to put into policy the basic rights that are going to allow the creators to control their creation, to charge what they want for their creation, and to develop the licensing deals that they feel they want to develop. And it’s gotten to be even bigger now than in the Napster days because back then, as you said, the transition was from CDs. So the artist only made money in those days during a few months after release, and then some CDs still sold for years afterwards, but a lot of them ended up in warehouses, and you kind of had this small window.

The streaming revolution allows “exploitation,” not in the negative sense, but in the positive sense. It allows it to last forever. That’s why you have all these new investors coming into the music community because it’s a little bit more like a commodity now. That’s why Goldman Sachs is buying catalogs. That’s why all of these new companies are coming in, and they’re buying catalogs of music because music’s evergreen. It’s water. People are going to love it forever.

And as an artist, you now have an opportunity you never had in the past. It’s not just to make this chunk on the wholesale cost of CDs within a certain month period; it’s, “Hey, now that I’ve built up this following and people might love these songs 20 years from now, guess what? I can now sell my catalog for a large amount of money to an investor in the future.”

I just want to hit pause here and say that the baby punk rocker in me is… We could do another full hour on artists who now have the opportunity to sell their catalog to Goldman Sachs. It’s another hour. I know it’s there.

Listen, we are all 20-something at some time, and then we’re 50-something at another time. Everybody wants to take care of their grandkids at some point.

But this is an opportunity that they have now that they didn’t have before. And so I think that this disruption… I think there’s more skin in the game for the actual creator than there was in the past when the revenue was more bifurcated or trifurcated. Nobody on the creative side of the equation thinks that it’s right that Suno will be making all the money from all of the music that has been created in the past, and that substitutional effect is going to deprive the artist or any of their partners.

If the artist is in the middle of the sun and you’ve got all the folks around them from performing rights organizations, to record companies, to publishers, to managers, to producers, to agents, et cetera, they’re all sort of partners in a piece of that larger puzzle. And now the artist has more skin in the game than they ever had before, which is another reason why I think we’re sort of rowing together a little bit longer.

There are other outcomes here. I think it’s Warner that’s partnered with Suno; they made a model on Warner’s music so they don’t run into these issues. If that proliferates, then you don’t get to sue the AI music companies because they made their models with licensed music, but you’ll face a supply problem on platforms like Spotify where you can suddenly flood the platforms with music.

Your solution there, at least as I understand this conversation, is labeling. You’re going to put labels on the AI stuff so the fans can make a market decision, and the bet is the fans will choose humans. Is there another solution to the glut of supply?

Yeah. I think an immediate solution for the marketplace for now is labeling. But I also think that these licensing agreements matter. I’m not privy to the licensing agreements that exist with Suno or any other AI company because I represent competitors, so they would never tell me. But I would imagine that it’s about dilution: where you don’t want to put artists’ music at risk on streaming platforms, because 90 percent of what they’re competing with is AI slop, if I could use that term. The licensing agreement gives you the leverage and the ability, if your rights are secured, to say, “I’m not going to let you download all this stuff.”

I mean, it’s rumored that in the Udio agreement, a walled garden had to be built so that fans could make music within the network on their own Spotify-like app. It wouldn’t discourage them from also having Spotify or Apple; it increases the revenue for everybody, but we’re not going to allow you to download from this platform and dilute that platform.

I don’t know if that’ll work. That seems to be one model. It could be that the terms with Suno had a limit on it or considered supply. I’m not a business person; I’m a policy person. But I think that there are plenty of ways, if the rights are secured, that through licensing arrangements and agreements, creators could then have the leverage to affect the supply so as not to cannibalize the work that they have on other platforms.

Well, I think the policy problem of supply is that it will affect the business somehow. We should talk about Spotify for a minute. It’s amazing that we haven’t spent more time talking about Spotify.

I know!

I think the model companies are not going to listen. You can see what they are doing. They are gluttonous for data, and they think that they can outrun the consequences. We’ll see what happens there. Spotify is not training a model. Spotify is just distributing music. That’s the thing that they’re doing; they’re a pure-play business in that way, and they have a lot of incentives for AI music to replace the music they have to otherwise pay for on that platform. They have a lot of incentives for white noise and fake streams in history to do these things, and they’ve gotten in trouble for some of it. This is a real thing that happens on this platform. Are they going to participate in a labeling scheme and say, “This is AI music?”

I think so, because as much as they have an economic incentive to lower their costs, their biggest economic incentive is to keep fans paying $9.99, $12.99, $15.99 per month. And so if fans feel duped, if you are putting —

Is that proven out in the market? I’m actually very curious about this.

Yes.

You’re saying there’s AI slop on Spotify now, or when you search for Taylor Swift, you get a bunch of weird Taylor Swift soundalikes, which is an experience everyone has on Spotify. It hasn’t churned those fans yet. There’s not some exodus to Tidal or Apple Music. Spotify is still dominant in its way.

Right. I think they’re walking down the road slowly, probably a little too slowly. I’m trying to see if they can do just enough to keep those two things in balance. Right now, for example, Spotify is saying, “We will only disclose or label a recording as AI in the credits.” Well, how many people look at the credits? I wish more people looked at the credits; they would discover producers and songwriters and other awesome folks who have contributed to the recording — but they don’t.

They need to do it at the track level. They’re not there yet. But they are disclosing AI personas that are made, because that’s sort of the highest on the dupe scale. And then under that, they’re looking at the fact that 90 percent of the recordings available on Spotify get 10 percent of the revenue, and 10 percent of the content gets 90 percent of the revenue. It’s because most of what’s on there gets less than 10 listens, or less than a hundred listens, and all of this slop is just going on, which means storage costs and all sorts of other things that they have to think about.

So I think that they will go as far down the road as fans will demand, as their users will demand. I think we’ll get to a point where people get these recommendations in their algorithm. “If you like this song, you’ll also like this one.” And when you listen to it, it turns out it was AI, you didn’t know it was AI, and you’ll feel deceived. That’s when they’ve got a giant problem because they’re responsible for that recommendation. They’re the ones that fed that to you. And we do have lots of consumer research out there showing that subscribers, just like all consumers, want to feel like nobody’s pulling the wool over their eyes.

But is there an actual market pressure? I can think of all kinds of times fans have revolted against Spotify. I love the band Stars, and every now and again they’ll tweet, “We didn’t make any money on this song on Spotify.” And everyone tweets at Spotify, and they’re mad. Spotify’s not paying artists any more money. Is there an actual market pressure that will make this labeling happen, or is it policy pressure?

It’s both. It’s a combination, because there are already laws in other countries that are mandating labels. And once a market like the EU or the US — that’s such a big market — or the UK or China, once they demand labeling on a global system, it’s actually cheaper to do it all around the world than it is to just geo-block it in one country.

This is the California car emission model of regulation, right?

It’s amazing what California can do because these companies are headquartered there. I mean not Spotify, but Apple and Google and others. And so the transparency law in California is the reason why Suno has to, as of August, put a watermark saying, “This is AI-generated” that’s machine-readable. One state, but they’re doing it all over the world because that’s what… So it’s a combination of policy, which obviously we are very directly engaged in, and competition. I predict that what Apple decides to do that is more fan-friendly will drive Spotify, and vice versa.

There’s not that much churn between them, but if there’s some sort of perception that you’re being more subscriber-friendly than somebody else, that tends to drive results. It does the same on the privacy issue, for example, where there’s no law on it, but now companies are competing to say, “We’re the private browser. Apple, we protect your privacy.”

One company says that. Let me ask you this. I’ve only got a few minutes left, and I want to take one turn at the end to talk about how you measure for platinum certification, because that’s also up in the air.

But let me just ask you one more question here on labeling. We have spent a lot of time at The Verge covering labels for images and media. There are all these standards that exist. I have one poor reporter, her name is Jess Weatherbed, and she has been forced to learn everything about the C2PA standard and content, SynthID, and all these things.

I’m so sorry. It’s really important, though.

They’re very important, and we spend a lot of time covering them because they have to work. And they have to work in a number of different places in the chain, right? They have to work in the camera, in Photoshop, on Twitter, and everyone has to agree. If Elon Musk does not want SynthID to show up on his platform, it doesn’t matter if it worked in the camera. This is a big chain.

It seems harder in music, much harder because a bunch of bedroom DJs are probably not going to participate in your technical scheme for labeling when they upload to DistroKid. An artist getting a beat from a producer over iMessage — who knows if that metadata will survive. You might have to watermark the sound files themselves in some deep intrinsic way, and that can get you into all kinds of trouble because now you’re screwing with the audio files at the software level and artists don’t like that. How do you make labeling work at all?

Yeah, you’re hitting on our daily meetings. So I mean, there are the international organizations like DDEX, for example, that can help align fields and stuff like that. C2PA, which can give you a model for provenance. But at the end of the day, it has to be the platforms and the distributors that are willing to do it. Because when it comes to generative AI, the record labels are not tagging and putting generative AI up on Spotify. That’s out of the DIY world.

That’s somebody going to a music generator, downloading it, and putting it up through DistroKid. So your gateway is DistroKid, TuneCore. It’s the distributors. Are they using AI detection technology to determine that it’s fully generated? You might not be able to use AI detection technology to say your song is five percent of this or use it for attribution yet, but you can definitely use it to say this is fully generated AI. And if so, are they willing to tag it so that it works with Apple and Spotify’s platform?

Then, on the platforms themselves, this is sort of like the parental advisory program; it might be voluntary, and everybody might have a little bit of a different way of implementing it, but fans are pretty clear. When something’s labeled as parental advisory, which probably just draws more young people to listen to it —

It did to me.

But when it’s labeled as parental advisory, people kind of get it. They know. It’s become sort of a culturally known thing. I think labeling for fully generated AI can definitely get there, but the distributors have to be willing to play. It can’t just be the platforms because if it doesn’t go downstream, they’re going to go in different directions. And already, Deezer’s going in one direction, Apple’s going in another direction, Spotify’s going in another direction. To me, the fully generated problem, which is a DIY problem, is a distributor problem. And we have to provide incentives to the distributors to get them to take responsibility and play ball.

Do you think that technology exists to support the policy there? You need a technology that can detect AI-generated media. I’m not sure that any of that exists or works well.

I think it exists and works well for fully generated AI music. We also do a lot of testing at the RIAA. That’s another thing that our tech team does. And then we disclose the results back to the folks who gave it to us. We won’t disclose it to anybody else, but if you want to know what our test says about your technology, we’ll tell you.

And I think when it comes to fully generated AI, there’s plenty of detection technologies out there that do a really good job. Once you get below that, time will make it better. So I don’t know that we could get into a lot of gradations right now. Deezer, very successfully, on a platform level, uses AI detection technology and then labels what it detects as fully generated AI, and it’s got a very low false positive rate.

I want to end by talking about the certifications for platinum records and so on. It’s an important part of what you do. All the artists want their little records on their wall. It’s amazing that it’s still records. One of our jokes at The Verge is also that no one knows how to count anything, that Nielsen ratings — I should probably disclose, we are now part of the company that owns Luminate, which powers the Billboard charts.

The Billboard charts and their methodology are always in flux. Actually, I think as of this year, they’re in a fight with YouTube. YouTube does not want to be counted on the Billboard charts because some methodologies are… Counting is very political.

It is.

You literally give awards for counting to a big number. That is in flux as artists gain more attention on TikTok. And I have probably heard that whole Nicole Kidman song, in fragments on TikTok. I’ve never listened to the whole thing all the way through, and I’m certainly never going to buy a copy of it. Again, all in flux, and then somewhere in there is a bunch of AI remixes, and now I’ve listened to “Pop Star” by Drake 10,000 times because Darth Vader is singing it to me.

How are you thinking about that methodology, as to who’s creating what? AI is in the mix, and we’re listening to more and more fragments to discover songs on social platforms than anything else.

I swear to God, you are in my daily meetings.

Can I just tell you? I’m happy to hear that. I drive The Verge team crazy because I’m like, the music industry is the future always. And they’re like, no one wants to listen to that. So thank you for indulging me. I appreciate you.

I mean, you have your finger on the pulse. This is exactly the type of stuff we’re thinking of. So right now we’re still in the world of 1500 streams equaling one unit, and what used to be the sale of a CD. Now it’s obviously singles-based more than albums-based, although we still do certify albums. And if you look at the chart eligibility program that is now being implemented outside of Billboard and OCC, in other places of the world, they have made rules to say, unless it’s substantially human, we are not going to allow it to be eligible for the chart because you’re literally taking the place of a human artist on number 49. Billboard, so far, will allow it, but they’ll label it.

We haven’t, because we certify albums that were released by record companies. We have yet to run into the issue of the AI slop or the machine-created piece. But as we get more into assistive AI, as creators start to use AI more, and you look at the chart eligibility of substantially human, does a stream as part of that 1500 mean that it still has to be a human stream? Yes, I’m sure that we will come down on that side.

But what does substantially human mean? I think over time that might change. It might be different three years from now than it is today. But at the core, it’s got to be in the control of the creator. And some people would argue that putting nine words into a music generator is an act of creation.

Oh, they will. They’re going to. Yeah.

And that they are a creator, and therefore it’s the same thing. It’s not the same thing. Not saying that craft doesn’t change over time, and that in songwriting it’s also different than in producing and creating a recording. Using AI in the songwriting process may look very different than it does in the recording process, so we have to think about that too. But as far as gold and platinum go, we are still in the reliably safe zone of being able to know with a great deal of certainty exactly what we’re doing. I think the evolution over time is going to be in how creators use AI more than it will need to be in something that’s fully machine-readable or machine-created.

Well, Mitch, you’ve given us so much time. As you can tell, I’ve thoroughly enjoyed this conversation. I could keep going for hours and hours now.

Me too. I mean, honestly, this has been awesome. You freaked me out a little bit because you… It’s so great to see, though, because the stuff that you’re focused on is exactly what we are trying to grapple with every single day. And I think where we’ve come to is that we can’t pretend that there are easy answers. We can’t just react sort of in a knee-jerk way. We’ve got to try to figure out a path forward that’s going to allow the industry to change and grow. This industry’s adapted more times than any other industry in the history of our country. I mean, nobody goes from wax cylinders to AI in a hundred years. Can you imagine asking Boeing to make cars? So, a very adaptable industry.

I agree with you that we’re ahead of the game and that we are the future. We’re sort of living it. And the lessons that we learned from the music industry, we can apply to a lot of other things. So this has been awesome.

Thank you so much for being on Decoder. Really appreciate it.

You got it. Take care.