This June, IFPI sprang into action. The trade organization for the recorded-music business around the world is best known for its annual Global Music Reports, which track annual economic growth and development for the industry by country and region.
But that month, it launched the first of three initiatives intended to tackle issues it sees as critical to the health of the global business: tagging and identifying AI-generated songs on streaming services; disqualifying AI-generated songs from its 60-plus international charts; and combating streaming fraud.
The measures were a relatively rare foray into consumer and business practices for the organization, which traditionally spends the majority of its efforts on government lobbying for legal protections for the creative industry and infrastructure support for the industry in developing regions. But for Victoria Oakley, who became CEO of IFPI in April 2024 after a decades-long career in British diplomacy and corporate consulting, the rapid advancement of technology, particularly AI, has made such industrywide consensus-building essential.
“I think music at this period in its evolution, with the challenges and opportunities that it’s facing, the real mix between tech, creativity, business and government, is a complex and really important blend right now,” says Oakley, who spent 18 years in the British Foreign Office before stints advising companies like Google, Apple and McDonald’s prior to IFPI. “I didn’t want to go to a trade body that argues about a press release that nobody really wants to issue. I wanted to come and do lobbying and advocacy and storytelling for an industry that I think matters.”
This year in particular, the music industry has confronted challenges that teeter on the existential, none more so than AI, which both offers huge creative potential and very real risks to the streaming business model and trust of consumers. And while the broader AI conversation is much bigger than music, Oakley is determined to make sure the sector doesn’t get lost in the shuffle. “In the regulatory and political world, I think music is a sector that’s beloved but not particularly thought about or understood, and that’s because we haven’t told our story,” Oakley says. “But we’re now at a time where AI is not coming; it’s here. And we need to be able to use AI as an opportunity for music, but also we need governments to help us build the right systems and guardrails so that music can continue to thrive in a Gen AI world.”
Since you took over, AI has been a major focus in many different ways. It is a gigantic topic, of course. What are some of your major concerns there?
I think having either this fear or this binary sense of what AI means for music or for the music industry isn’t very helpful. I think what’s quite impressive about music is we’ve got to grips with AI quite quickly. [The record industry] actually has figured it out, knows a lot about what it wants to do with AI and how it wants to use it. But also how [music] needs to be protected, and how the way that the industry and the ecosystem works — and [the way] that the money moves around in order to have a thriving, sustainable industry —needs to be protected. So, kudos to leaders of record labels for getting to grips with it, moving toward licensing.
Related
The concern is there are too many conversations still in governments that don’t particularly understand the music ecosystem and how it works, that are proposing legislation that they think will be fine, but is actually throwing creativity entirely under the bus. The most live example of that is in Australia, where we have a government that appears to be entirely listening to the promises and the threats of a number of AI companies, saying you absolutely have to throw copyright out, which essentially means throw out the basis for how creativity can work and people across the creative sector, not just music, can make their livelihood from their creativity. “You have to throw this away; it doesn’t work. We can’t possibly come and invest in your country unless you do this.” And it’s ludicrous. We all know there are solutions. We’ve all seen lots of governments grapple with this and either find a way through or step away from the exceptions that Australia is considering. So that is my worry: that governments risk jeopardizing their own creative sectors and culture, and all that can come from that that matters to us as human beings.
How did you develop the principles for your streaming fraud initiative, and what are you hoping to accomplish?
We start from the principle that streaming fraud is dumb and bad and damaging everyone involved, whether you are an artist or a fan or a record label or, frankly, a DSP. Right? It’s just rubbish. So, what are the ways that we can fix that? If there was one single easy answer to this, then we’d have fixed it ages ago because we’d all yell at the one guy until he fixed it, and there we go.
But the point is that there is a role for multiple stakeholders across the streaming value chain. So what we’ve tried to do is identify what are the measures that all of those players could and should put in place. Some of which are quite basic: know-your-customer type things. Some are about good vetting, and then some are about really good data sharing, because we think there’s a whack-a-mole problem here as well, where fraudsters rock up somewhere, you make it difficult there and they go somewhere else. So our strategy is really about increasing friction and making it more difficult for fraudsters to succeed.
How do you do that?
I think it’s too easy to make money from streaming fraud. So if you increase friction, they will say, “Oh, this is hard work now, not for me, thanks.” And so we’ve been looking at all of the places where you could make it more difficult. Whether that is about accessing a distributor or platform in the first place, securing the revenue or being able to jump easily from one place to another and not hide, as it were. So that’s how we designed the principles across record labels, really saying what do we think we can see in the system which causes the problems.
What we’ve moved into now, though, is trying to road test that with others. So our first wave is with distributors, and I don’t want to simply tell them what to do. We also want to have a conversation with them about what they think. Most distributors don’t want streaming fraud in their business, either. So a healthy exchange that says, have we come up with the right things? Are these the right principles? What else? What more? What different could we do? And then, how do we move toward compliance where more and more players are living up to those principles and ensuring that they’re applying them, but also then continuing the conversation. What more might we need to do?
How does this differ from or dovetail with other anti-fraud initiatives that the industry has rolled out in the last couple of years?
This one is an attempt to say, “Here are a set of measures that all of us could and should be taking.” I haven’t seen anything quite like that before that has this sort of collaborative approach. We’ve been in conversation with the [Music Fights Fraud Alliance] throughout, which is the organization set up a few years ago, to try and have this conversation. They have been looking at a number of measures, and they’ve been looking at some things that we haven’t looked at, like artist education about not accidentally falling prey to some of these services that you think are going to benefit your career, but are actually offering fraud, and we’re not particularly working in that space. This is quite focused on very practical steps that distributors could take. There is a piece of work to come on what other practical steps that DSPs could take. We’ve started with the distributor piece because that is the gateway in, but this doesn’t all sit with distributors alone.
What role do DSPs have in this?
I think there’s some interesting stuff about thresholds and volume and identification of a single economic actor, for example. I think DSPs are doing lots of things, and they’re looking for different red flags that might imply anomalies that merit investigation, although not automatically fraud. Are we all applying the same flags? Are we all defining the threshold in the same place? Are there other things that we could collectively look out for? Is there a case for doing any of this more collectively or in a more standardized fashion? We’re still working on what we think that looks like, and our intention is to have the same sort of approach. Inviting DSPs to join us on that journey, have that conversation with us and then at some point linking the two, that will make a difference. And then making sure those are aligned, because what I’ve experienced until now in the industry is a bit of finger-pointing. The DSPs point at the distributors, the distributors say it’s the DSPs. Then it’s, “If only the record labels did this” or, “Actually, it’s that kind of distributor versus this kind of distributor.” And as long as we’re all blaming one another, we’re allowing the fraudsters to have a free ride because we’re busy shouting at ourselves.
Classic music industry.
Not the only industry.
Are you working to ensure compliance, or is this more about recommendations on how to police yourselves?
We’d like to move toward a form of compliance. I think that’s super important, and I think I would almost like accountability more than compliance. They’re only slightly different concepts, but it’s an important difference in tone. You could have a centralized audit function where somebody is acting as the global policeman, and you can only be considered compliant and have the badge if you adhere to all of these things. I don’t think that’s IFPI’s role for now. Nor do I want it to just be a piece of paper where everybody says, “That’s marvelous.” So we think there’s something in the middle that comes partly from this collaborative process and the conversation that’s probably about self-audit, but probably a pre-agreed, systematized self-audit.
Related
We think this exchange of views and trying to improve together and come up with the other things that we could all be doing is probably more useful than yelling at the person who’s made the least progress so far. But let’s see. I think we are on a journey to compliance, so I don’t think we just leave it as a kind of lovely, fluffy announcement out in the air. But I’m not yet at, “Vickie Oakley is the global policeman and gives you a sticker because you’ve done all your homework.”
You mentioned the whack-a-mole issue, which I think is a very big component of this. But also the announcement did not include companies like TuneCore and DistroKid and OneRPM. Was there a reason that some of those DIY, single-fee upload distributors were not included in the announcement?
We made the announcement with a number of distributors that we knew and were working with, and who’d come on board to help us shape it. I’ve since had conversations with around 30 more distributors. My phone was ringing off the hook, metaphorically, with distributors who wanted to understand it, add things to it, offer a view, get on board, all of that. So we are now working our way through talking to literally any distributor who’s interested, whatever size or shape.
Look, there are good actors and bad actors out there, right? And what I’m glad about is to be in the conversation, raising awareness, raising the temperature on this, so that we can get to a point where everybody is wanting to comply as best they possibly can to stamp this out.
You guys also rolled out new AI labeling proposals, where songs will be marked as fully AI-generated or AI-assisted. What was behind that initiative?
I would like to live in a world where, as a consumer and a music fan, I am informed pretty easily because I am quite a lazy consumer, and I long for a world in which it’s very easy for me to experience my music and know what I’m getting and what I’m not getting. That is our starting point. So to try and make that easier in this very complex world, we came up with this proposal, which is a scheme that says we should label everything, because although we announced two labels, there’s essentially three. There is fully AI-generated: you know, if your product was made entirely with an AI tool. Then there is AI-assisted, which admittedly is much more complicated. And then there is no label, which by definition means either zero AI involvement or very, very minimal.
Related
I think that’s probably right for now. It may well have to change and evolve in the future, but I believe that consumers want and merit a pretty simple piece of information wherever they are in the world. I know that isn’t incredibly straightforward, but it’s also not incredibly complicated to do. We will start to have all of that built into the metadata and the plumbing that comes from artists or labels, and that will work its way through distributors to DSPs. But DSPs are the ones who have the relationship with the consumer. That’s where the magic happens, and so that’s where the label needs to be easy to find, easy to see, easy to understand.
Now, I might say I think this music is absolutely the best thing I’ve ever heard, and I’m thoroughly informed now that it’s AI and I still want to continue to listen to it. And actually, this is what I like to listen to in the shower, in the car. But I know who it is and what it is, and I also know that I probably can’t buy tickets to go and see them live. So I already know that I’m in a slightly different relationship with that music, and that’s fine because I’m going into it with my eyes open. So we are in the midst of conversations with DSPs about what adopting this could look like, because I think there’s a real patchwork out there at the moment. I don’t think it needs to be in big flashing red lights, but it should be consistent and readily available to me as a consumer to make an informed choice.
You’ve said it would involve voluntary disclosure of AI usage. How difficult would it be to actually have this as a uniform label?
You’re right — what we announced earlier in the year is a voluntary scheme, and I think there are a number of reasons for that. One is, this is for others to carry out, and we are super open to a conversation with all of the platforms if there are things about it that they find we could improve or should be better or could be clearer. The other thing is, you have got a number of jurisdictions in the world where there either already is a requirement to label, certainly in the EU and China and a whole bunch of countries that are considering one right now. And then I suspect more countries will come into that and join that party. So one of the things we wanted to do was set out what we think is the best industry standard, so that ideally platforms will adopt it and use it, and then there’s no need for legislation, arguably. But where legislation is being considered, they could take this voluntary scheme as the absolute model for it.
Personally, I’d like to see some real energy and movement behind this because we do have this weird patchwork that is inconsistent for consumers, and the conversations that I’m about to start having are, “Can we all get on board and do this together?” And if we need to tweak it and change it because there’s a bit about it you don’t like or that doesn’t work, then let’s do that. But let’s offer consumers what I think they merit and need.
But what would the incentive be for an artist or a label to even disclose it, especially for an AI-assisted track, which we know covers such a wide spectrum?
I think artists, as a general rule, are pretty prepared to be open about the tools that have led to their creativity. I don’t think anybody should be ashamed or afraid to either use AI or admit that they’re using AI in developing their craft. There is currently no disincentive, right? You’re not going to not get playlisted. You’re not going to have Spotify not have your tracks on their platform. I think you’re being transparent with your fans and with consumers. So I hear you, but I’m unconvinced by the argument that this is risk-heavy for artists.
There was also a third initiative regarding labeling AI tracks on the charts. How should that work?
What we fundamentally have always believed is that the IFPI charts, of which there are over 60, are a way of celebrating human artistry and measuring the market and what’s working and what’s not. So we have argued that if a track is fully AI-generated or made through an unlicensed model, then they do not merit their place in the chart, which is there to measure human artistry that has been made ethically.
So there are a number of tracks being taken out of the chart, and we’re not saying this track is not popular or this track is not good. We’re saying whether it’s because it’s fully AI-generated, or it’s been made using an unlicensed model that therefore is by definition not ethically trained, then it doesn’t merit its place in these charts. So far, it’s working. There are people who don’t love it, and that’s OK. Do I think one day there might need to be multiple charts, and one that measures fully generated AI tracks against other fully generated AI tracks? I don’t know, maybe, but for now I do not want to see works that were made with unlicensed models, i.e., stealing other people’s work by definition to get to their output, be celebrated in the same way as a track that an artist has put their heart and soul into, sit there in the same chart.
I also want to make sure that we’re comparing apples and pears. Let’s compare how well this piece of human-created work is doing against this piece of human-created work. Again, it’s about ensuring that music fans have transparency and knowledge, and if we feel the need one day in the future where there is more and more licensing in place, and there are good, agreed, ethically trained AI music models that are producing things, then let’s revisit how we measure those at that point. And there is a place for that music, but let’s be very clear about what it is and what it isn’t.
Setting AI aside, if we can, what other areas are you focused on with the organization?
AI is pretty dominant, and rightly so, because it is opportunity and risk, so it would be wrong if we were not spending the bulk of our time on that. To be honest, streaming fraud comes very close because it’s such a dumb loss of money from authentic artists to bots or bad guys, and I would like that to stop.
We then spend a lot of time trying to make sure that across the globe, the story of music and the machinery behind music is really understood. And particularly if you think about the parts of the world that we’re focusing on — Latin America, the fastest-growing region for a few years in a row; Asia going great guns; different things happening in different parts of the world. China has seen absolutely exponential growth, off the scale. For the second year in a row, Vietnam is actually the fastest-growing market in the world. You’ve got these incredibly interesting markets developing at very different paces, but some of them very, very fast. The 20% growth rates for some of these countries is phenomenally impressive, and yet they are all at different stages in their development of their licensing market, their collecting society arrangements, how and whether they do synch. Whether or not in India you can charge for when music is played at a major event, like a wedding. So there are all these bits and pieces of how the music industry continues to make revenue for its artists and to be heard when you step outside of the Anglo-Saxon mold that all too often we spend our time in.
I did want to ask about two of those fastest-growing markets from the latest IFPI global music report: China and Brazil. What do you think is behind the particular growth there?
China is coming into its own, and where it should be, right? And I think, reflecting on India, both of them are countries with huge musical talent, huge music consumption and huge populations. To some extent, India has finally taken its place in the last few years on the global stage in the way it should. The other thing is the Chinese government has been very clear about their determination to make music and their cultural ecosystem an important part of their economy. They have impressive homegrown DSPs, impressive homegrown record labels, phenomenal homegrown talent that at the moment isn’t exported hugely, and that will be really interesting to see. And like in many parts of the world, there’s huge domestic consumption, so they listen to their own music. But it’s a government with real intent behind what it’s put around music. It’s been pretty clear on AI, it’s got AI labeling ahead of a lot of other countries. They’re doing a lot to stamp out piracy, having perhaps a decade ago been on the top of everybody’s list for piracy.
Brazil is a slightly different story. I think we’re seeing the music economy emerge at the same time as we’re seeing the Brazilian economy emerge. It’s also a place of phenomenal talent, and their export story is much more successful so far, principally across their own region. But I think given that most of the music is in Portuguese and not in Spanish, so the obvious sellable market isn’t there in quite the same way, they’ve done a phenomenal job with their export story and they are on the increase. They too have a government that I think, from early on, places culture as an important part of the education system, and I think that is quite an important first step that we often don’t see the benefits of until much later.
The biggest thing, though, is talent. You can have the greatest systems in the world and perfect licensing rules and phenomenal public performance laws, and everybody’s on board and nobody’s corrupt and everything’s brilliant — but you have to have musical talent that is encouraged, first and foremost.
As the rate of streaming revenue growth slows, how does the industry overall continue to grow?
Slows where? Because I think there are parts of the world where it’s slowing, and those are definitely among the most significant markets. But there are parts of the world where it’s growing very fast, and there are parts of the world where it still has room to grow. So I think what we have to focus on is that potential for growth, even in quite established markets like France and Germany. Somewhere like the U.S. is a pretty saturated market already, so by definition, there is less room for growth. You look at somewhere like India, where there is phenomenal room for growth because we are streaming, but we don’t have a lot of paid streaming. So I think it’s about a diversification of where that streaming revenue comes from, and being very clear about which markets have that potential for growth and understanding what unlocks it.
What are you looking forward to in the next six, 12 months that you hope to accomplish at IFPI?
We are really trying to, on the record label side, put forward views on positions for how we make all of this better for everyone and work with other stakeholders, DSPs, distributors, DIY artists, the whole spectrum, and I think that collaborative process is important and healthy. I don’t know if it was there before. It feels refreshing to me, but it also feels like a drive to everybody upping their game, whether that’s on AI labeling and transparency for consumers, whether that’s on what should be counted in a chart and what shouldn’t be as true human artistry, or how we clamp down on streaming fraud effectively.
I think we’ve got work to do, and I think the DSPs and the distributors have got work to do, and we’ve been pretty clear about that and our desire to work with them to achieve a better, fairer system. Because there are places to hide, and I’d rather there weren’t, so that we can all have a music business that’s still thriving 10 years from now, and where one of my kids could say, “I want to be a musician” and my heart doesn’t sink.
