Following a Song — The Tape

On 14 April 2026, an Australian musician named Jo Loewenthal filed suit in the NSW Supreme Court against Universal Music Publishing Pty Limited. He has spent the past several years compiling evidence about how royalties for his band Tora’s music move through the global music rights system, and what he found is the reason he is now in court.

The numbers, as he has made them public, are unflattering. A SOCAN-APRA email exchange he obtained through Canada’s privacy laws shows the two collecting societies discussing how to redirect his data requests in circles. SESAC, an American performance rights organisation, went 53 days without acknowledging his formal data access request and then refused to disclose, citing a privacy exemption that operates on behalf of broadcasters. The royalty file APRA AMCOS eventually gave him, after he invoked the Australian Privacy Act, contained 11,000 rows of seven years of transactions across Spotify, Apple Music, Pandora, YouTube, and Tidal. It contained no per-stream play count anywhere.

Loewenthal is now rallying other artists under a banner referencing GDPR Article 15, the data access right that lets EU citizens demand their personal data from any company holding it. He is asking artists to file their own access requests. He is asking the ACCC to refuse APRA AMCOS its current bid to renew its blanket-licensing authorisation. He has called the system, in correspondence the regulator’s own case officer agreed with internally, “quite possibly fraud and/or incompetence.”

We are not writing this to argue his case. That is what the NSW Supreme Court is for. We are writing because the campaign Loewenthal is running has done something more useful than win or lose a single legal action. It has made visible the part of the music industry that working artists have, for the better part of a century, been told to trust without examining: the chain of organisations and contracts and statements and pools that converts a song being listened to into a payment showing up in someone’s account.

Most working artists have never seen that chain laid out clearly. Most of the time, that’s because nobody in the chain has any particular interest in laying it out. The industry has accumulated complexity for a hundred years, layer over layer, contract over contract, society over society, and arrived at a system that even its own auditors describe as opaque. Some of that complexity is necessary. Music has always been a layered set of rights. A song is a composition, and a recording, and a performance, and at any given moment three different organisations might be collecting money on behalf of three different rightsholders for three different uses of it. The complexity itself is real.

[ Pull-quote candidate ] But complexity that benefits the people inside it, and that the people outside it cannot inspect, is not a neutral fact. It is the place where the system’s failures live, and it is the place where most of an artist’s potential income quietly disappears.

This piece is an attempt to lay the chain out plainly. How a song earns. Who collects what, on whose behalf. Where the gaps are. What an artist can see, and what they cannot, and why. We will use Loewenthal’s evidence where it makes the abstract concrete. We will use international examples because the music industry is international, and because the same patterns appear in different jurisdictions with different regulators and different names. We will not take a position on what the system should look like. We will describe what it looks like now, and where the holes are, and let you judge.

The article is long. It is long because the system is a maze and a maze does not get clearer by being summarised. We have included four interactive elements to help. Read in order or jump to the section you need.

The basic split

Every song generates two separate copyrights. The composition, which is the song itself: the melody, the chords, the lyrics, the structure as written. And the recording, which is a particular performance of the composition committed to tape, or to digital, or to vinyl. These two copyrights are owned, licensed, and paid separately. They are also, in most cases, owned by different people.

If you wrote and recorded a song alone in your bedroom, you own both. As soon as you brought in a producer, an engineer, a co-writer, a session musician, or a label, you began carving the two copyrights up between yourself and other parties. Most of the disputes and most of the missing royalties in the music industry happen at the seams between these splits. Knowing the split exists, and how it sits across the people who worked on the record, is the most useful single fact a working artist can have.

Every time the recorded version of your song is used somewhere in the world, it generates revenue under as many as three different categories.

There are mechanical rights. These are paid for the act of reproducing the composition. Every download, every CD pressed, every vinyl cut. Streams also generate a mechanical component, splitting the per-stream payout between the performance side and the mechanical side. Mechanical royalties are owed to the songwriter and the publisher: the people who own the composition.

There are performance rights. These are paid when the composition is performed publicly. Every radio play, every sync placement on a TV show, every streaming play, every live performance in a venue licensed for music. Performance royalties are also owed to the songwriter and the publisher, but they are collected through a different organisation than mechanicals. In Australia, that is APRA. In the UK, PRS for Music. In the US, ASCAP, BMI, SESAC, or GMR depending on which one the songwriter is signed to.

And then there are neighbouring rights, which most artists do not realise exist. Neighbouring rights are paid to the performer and the master rights holder for the public performance of the recording, separately from the underlying composition. When a record gets played on the radio, the songwriter side gets a performance royalty, and the recording side gets a neighbouring rights payment. They are two different organisations, two different statements, two different audit trails. In Australia, neighbouring rights are collected by the PPCA. In the UK, by PPL. In the US digital-only, by SoundExchange. In most other countries, by separate societies again. Loewenthal’s accusation against The Orchard is a neighbouring rights accusation: he is alleging the distributor has been collecting Tora’s neighbouring rights since 2017 without authorisation.

This is the architecture under everything else this piece will cover. A composition has a publisher and a songwriter. A recording has a performer and a master rights holder. Each pair has a collection society for performance use, a collection society for mechanical use, and a collection society for neighbouring rights. That is a minimum of six organisations involved in a single song’s earning, every time it gets used. Most working artists know two of those six exist. The rest of the chain operates without their direct visibility.

The creation chain

A song does not arrive at a streaming service with its rights pre-arranged. It arrives at the end of a chain of decisions, contracts, splits, and registrations, most of which happen quietly during the recording process and almost none of which the average listener will ever see. Knowing the chain matters because every link of it is where money either flows correctly or, more often, does not.

Start at the writing.

Most modern songs have multiple writers. A producer who built the beat. A topliner who wrote the melody. A vocalist who wrote the lyrics. A featured artist whose verse came in late. Each of them, by default, is a co-author of the composition, and each of them owns a share of it. Those shares are negotiated, sometimes through written split sheets signed in the studio, more often in the gap between “we should write this down” and “we’ll sort it out later.” A song that becomes successful with the splits unwritten generates a problem that gets harder to solve every year it earns money.

Each songwriter then decides what to do with their share. Self-published writers register their share directly with their PRO and a mechanical-rights body. Most working writers eventually sign with a publisher: a company that handles the registration, the chasing, the licensing, and the auditing in exchange for a percentage of the income, typically 20 to 50% of publishing income depending on the deal. The publisher’s share is sometimes called the “publisher’s share” of the composition; the writer keeps the rest, often called the “writer’s share.” The two move through different parts of the collection chain even though they originate from the same registration.

Now turn to the recording side.

The recording is a separate copyright. It belongs, by default, to whoever paid for the session: the artist if they self-funded, the label if they signed a deal. A standard major-label recording contract grants the label the master copyright outright. A typical “label services” deal lets the artist retain the master in exchange for a higher cut to the label. Independent artists who record on their own are the master rights holders by default.

The performers on the recording, separately again, have neighbouring rights. A guest vocalist who sang on the record is paid neighbouring rights when it gets played publicly, even if she does not own a share of either the master or the composition. These payments flow through a different society again: PPCA in Australia, PPL in the UK, SoundExchange in the US for digital broadcasts only. Most session performers do not register and never receive these payments. The money sits in suspense accounts until it is forfeited or claimed.

The recording, with its master rights and its neighbouring rights, then needs to get distributed. Either through a major label that owns the master and handles distribution as part of the deal, or through an independent distributor, or through both: a label-services deal where the label owns or controls some rights but the artist retains others.

Distributors do two things. They get the recording to streaming services and digital storefronts, and they collect the money those services pay back. That sounds simple. It is not. Distributors take a fee for the work, and the fee structure varies wildly. Flat-rate distributors like DistroKid charge a yearly subscription, around US$24.99 for unlimited releases, and keep zero of the streaming revenue. Commission distributors like CD Baby take a percentage of the income forever, around 9%. Higher-touch distributors like AWAL take 15% in exchange for label-services support. Pure label-services agreements can run anywhere from 30 to 50% but bring marketing, sync teams, and other promotional infrastructure.

The distributor passes the money to the master rights holder, which is sometimes the artist, sometimes a label, sometimes both. If a label is in the chain, the label takes its cut: typically 75 to 85% of master royalties on a traditional major-label deal, with the artist keeping the remaining 15 to 25%. Newer artist-friendly deals can land closer to 60/40 or 50/50, and any deal involving recoupable advances comes with the artist’s share withheld until the advance is paid back. After all of that, the artist’s manager takes around 15% of what reaches the artist, and the lawyer takes a cut of any deal they negotiated.

A track on a traditional major-label deal that earns US$30 in streaming royalties can, in practice, leave around US$2 to US$3 in the artist’s bank account by the time it works through this chain. An independent artist using a flat-rate distributor with no label, no manager, and no lawyer, and who has properly registered both their composition and their recording, can keep substantially more, sometimes US$20 of that same notional dollar after distributor fees. The variable that determines which version applies is not how many people listened. It is how the chain of contracts above the song is structured.

The widget below lays out each link of the chain, who collects what, what the typical splits look like, and where data is, or is not, visible to the artist.

From song creation through to listeners

DSP economics

When you press play on a song on Spotify, that play does not generate a fixed payment to the artist. It generates a fraction of a payment, and the size of that fraction is determined by a pool calculation that runs across the entire platform every month.

This is the part of the streaming economy that surprises most artists when they learn it.

The standard model is called pro-rata, and every major streaming service uses it as their default. Each month, the platform takes its total revenue from all subscribers and all advertising, sets aside its own cut (around 30% on Spotify, similar elsewhere), and pools the remaining 70% as the rights-holder revenue for that month. That pool is then distributed to all rights holders proportionally to their share of the platform’s total streams that month.

The math has a counterintuitive consequence. If you stream one of your favourite indie artists every day for a month, and the listener next to you streams nothing but Taylor Swift, your subscription dollars do not flow to the indie artist. They flow into the same pool that covers Swift’s streams, and Swift’s share of total platform streams is many orders of magnitude larger than the indie artist’s. Your A$13.99 subscription is functionally a vote in favour of whichever artist commands the most plays globally that month, regardless of what you listened to.

The alternative, user-centric, divides each subscriber’s monthly fee only between the artists that subscriber streamed. SoundCloud has been running this model for indie artists since 2021. MiDIA Research found 56% of artists on the system better off than they would be under pro-rata; SoundCloud claims indie artists earn around 60% more under user-centric. Spotify, Apple Music, Amazon Music and the rest have not adopted it. Deezer, with Universal, developed a hybrid system they call “artist-centric” that caps each user’s contribution to any single artist at 1,000 streams a month and double-pays “professional” artists, but it is not the same as user-centric, and the major labels broadly back it instead.

The widget below lets you see the difference for yourself. The same A$13.99 subscription, the same listening, two different models, two different outcomes for the artist you actually streamed.

Pool model vs Direct pay

Inside the pro-rata model, DSPs have built a series of additional rules that further reshape who gets paid and who does not.

The most consequential is Spotify’s 1,000-stream threshold. Since April 2024, a track must accumulate at least 1,000 plays in any rolling 12-month window to earn recording royalties at all. Below that threshold, the track stays on the platform, the streams keep counting publicly, but no royalty is paid to the rights holder. Spotify says 86% of all tracks on the platform sit below the threshold and account for only 0.5% of total streams. The royalties that would have been paid to those tracks, around US$40m in 2024 alone, do not leave the platform. They get redistributed proportionally to tracks above the threshold, which means they overwhelmingly flow to the artists already streaming most. Critics, including Disc Makers, framed the same number as roughly US$47m withheld from the smallest indie artists. Both are true descriptions of the same money, just with different lenses on who it should have gone to.

The threshold sits alongside a 2-minute minimum listen requirement for noise and functional content to qualify, and a €10/track/month penalty Spotify charges to labels and distributors for tracks flagged for high artificial-streaming activity. The penalty is meant to deter bot streaming, which is real and rampant. Spotify removed around 75 million tracks for fraud and spam over the 12 months ending September 2025, and a 2024 federal indictment in the United States described a single operator who used AI-generated tracks and streaming bots to extract more than US$10 million from Spotify, Apple Music and Amazon Music. But the penalty also catches legitimate artists with viral spikes or playlist activity that the algorithm flags as suspicious, and distributors typically pass the penalty on to the artist regardless of cause.

Then there is Discovery Mode. Artists or labels can opt selected tracks into Spotify’s Radio and Autoplay placement in exchange for a 30% commission on the royalties from any qualifying streams. Promotional placement, in other words, in exchange for a percentage of the income from the resulting plays. The Recording Academy, the American Association of Independent Music, and a number of working artists have called the structure digital payola. Spotify’s framing is that participation is opt-in and rate-protected. Both descriptions are accurate. The disagreement is over what to call it.

What an artist earns per stream, after all of these layers, varies by service and varies considerably by listener territory and subscription tier. The blended averages reported by the music industry trade press in 2025 are these. Tidal pays the highest, around US$0.012 to US$0.015 per stream. Apple Music pays around US$0.007 to US$0.010 per stream and is consistent in returning around 52% of its revenue to rights holders. Amazon Music pays roughly US$0.004 to US$0.008. Spotify pays around US$0.003 to US$0.005 and allocates around 65 to 70% of its revenue to the rights-holder side. YouTube Music, which mixes ad-supported and paid streams in a single bucket, pays the least, between US$0.0007 and US$0.002 per stream.

These figures are blended averages, not contractual rates, and any artist quoting them as fixed rates is misreading them. But the orders of magnitude are stable across reporting cycles, and the picture they sketch is the same picture every working artist has been describing for years. A track with 100,000 streams a month on Spotify, after the rights-holder side pays its labels and distributors and PROs and publishers, is, for most artists, not a working income.

The widget below ranks the major services by their average per-1,000-stream payout. Distrosub appears separately because its model does not pay per stream at all.

[ HTML WIDGET — Widget 3: How streaming services pay. Paste contents of widget_03_dsp_payouts.html ]

How streaming services pay

Below is the breakdown of where music listeners are paying their money, by region and currency. The category share is not the artist’s share. It is the share of total industry revenue. The chain between the listener’s subscription and the artist’s monthly statement passes through every layer this section has just described.

[ HTML WIDGET — Widget 2: Where listeners are. Paste contents of widget_02_where_listeners_are.html ]

here listeners are

PROs and the matching problem

PROs sit one layer below the streaming services in the chain we have been building. They handle the songwriter side of every play. Every time a song with registered composers gets used publicly, anywhere, the PRO is supposed to identify the song, identify the rightsholders, and collect the royalty owed.

There are dozens of PROs globally, organised by territory. APRA AMCOS is the Australian one. PRS for Music handles the UK. ASCAP, BMI, SESAC and GMR cover the US, with songwriters typically signed to one. GEMA is German. SACEM is French. JASRAC is Japanese. SOCAN is Canadian. Most countries have at least one. They coordinate internationally through CISAC, the Confédération Internationale des Sociétés d’Auteurs et Compositeurs, an umbrella body that links 227 member societies across 116 countries. CISAC reports the global PRO network collected €13.09bn in 2023 alone.

The job, on paper, is clean. A song plays on a radio station in Berlin. GEMA logs the play, identifies the song and writers, takes its administrative cut, and either pays the writers directly (if they are GEMA members) or routes the money to the writers’ home PRO via reciprocal agreement (if they are not). A Japanese citizen’s song getting played on Australian radio gets collected by APRA, sent to JASRAC under the reciprocal arrangement, and paid to the writer through their home society. The international web of reciprocal agreements is what allows a writer in one country to be paid for plays in another.

The reality is messier than the structure suggests, in three specific ways.

First, the matching of plays to rightsholders fails routinely. PROs use a combination of direct usage data from broadcasters and digital services, sample data, music-recognition technology, and proxy data where direct collection is not cost-effective. APRA AMCOS alone reports processing around 6 billion music-usage records annually. At that volume, mismatches are inevitable: a play attributed to the wrong song, a song attributed to the wrong writer, a writer’s name misspelled in the upstream feed and never reconciled. Matched royalties get paid out. Unmatched royalties go into what the industry calls black-box accounts, where they sit until they are claimed or, after a period that varies by jurisdiction, redistributed to other rightsholders proportionally based on market share. The black-box redistribution mechanism, in practice, sends most of those unclaimed funds to the largest rightsholders.

Second, the data made available to artists about their own plays varies by PRO and is, in most cases, well below what the PRO itself holds. A typical PRO statement gives a member their distributions per work per quarter, broken into broad categories of source (broadcast, live, streaming, sync), but with no per-stream or per-broadcast detail. Members can in many cases dispute or audit specific entries, but only if they know what to dispute, and the data they receive is rarely sufficient to identify a specific anomaly without significant forensic effort.

Third, the international web of reciprocal agreements introduces additional opacity. When money moves from one PRO to another under reciprocal terms, the receiving society applies its own administrative deduction, its own distribution timetable, and its own categorisation before passing the money to the writer. By the time a payment reaches a writer in Sydney for a play in Hamburg, several deductions have happened, several conversion rates have been applied, several reconciliation cycles have passed, and the writer has visibility into none of those steps.

This is the system Loewenthal has been documenting. The 11,000-row data file SESAC produced for him, after a 53-day delay and only after he filed complaints with the FTC and the Bavarian data protection authority, is the artifact of this layer. It contains seven years of transactions across multiple platforms with no per-stream usage figure attached. The internal email correspondence between APRA and SOCAN, obtained via Canada’s PIPEDA legislation, shows two PROs coordinating across two countries on how to respond to one artist’s request to see his own data. The forensic report he submitted to APRA AMCOS, identifying 111,977 exact duplicate rows and hundreds of unexplained negative royalty entries in the data they had given him, returned the response: “this is the best we can do.”

The pattern is not unique to Australia or to Loewenthal. The same structural design, the same data thinness, the same coordination through CISAC, operates in every major market. The names on the statements change. The categories under which royalties are summed change. The fundamental arrangement, where artists are sent a number and asked to trust it, does not.

In the United States, the equivalent layer for songwriter mechanical royalties is partially handled by the Mechanical Licensing Collective, established under the 2018 Music Modernization Act and operational since January 2021. The MLC has distributed more than US$2.5 billion cumulatively as of late 2024, but roughly 50% of black-box royalties remain undistributed more than three years after launch. A 2025 investigation by Digital Music News flagged hundreds of suspicious works exploiting duplicate-metadata gaps, suggesting the same matching failures the international PRO network has dealt with for decades have not been solved by regulatory restructuring alone. The MLC has also sued Spotify over the 2024 audiobook bundling reclassification that allegedly cut publisher royalties, and faces ongoing governance scrutiny: every voting publisher member of the MLC board in 2023 and 2024 was also on the board of the National Music Publishers’ Association.

The thing that makes the matching problem hard to fix, in any jurisdiction, is that the institutions doing the matching are the ones being paid to do it. The redistribution rules for unmatched royalties send them to the largest rightsholders in most major systems, which means the institutions with the most to gain from imperfect matching are the institutions in charge of improving it. Independent reform proposals, including the “Direct Programmable Licensing” model Loewenthal advocates, have struggled to find institutional sponsors for exactly this reason. The current system’s failures benefit too many of its operators for those operators to lead the redesign.

[ Pull-quote candidate ] The system is not an accident. It is the equilibrium state of a hundred years of layered contracts and reciprocal agreements between institutions that, structurally, have more to lose than to gain by improving its transparency.

What the artist actually keeps

Section three walked the chain from songwriting through publishing, recording, and distribution. This section pulls one specific knot in that chain into focus, because it is the knot working artists have the most direct control over: which distributor they use, and what that distributor takes.

Most independent artists in 2026 sit on one of a small set of distributors. DistroKid, the largest by volume, charges a flat US$24.99 a year for unlimited releases and keeps zero percent of streaming revenue on its base plan. TuneCore charges tiered annual plans, currently US$14.99 for the Rising Artist tier, US$29.99 for Breakout, and US$49.99 for Professional. CD Baby, an older operator, charges US$9.99 per single or US$29 per album as a one-off and then takes a 9% commission on royalties forever, in exchange for keeping the catalog up indefinitely with no annual fee. AWAL, owned by Sony, is invitation-only at the upper tier and takes 15% in exchange for label-services support, with a starter tier around US$19.99 a year. Stem and Symphonic sit in the same neighbourhood, each with different mixes of flat fees and commission. Amuse offers a freemium model. EMPIRE works closer to a label-services agreement with negotiated splits.

The fee structure changes the math at scale. A flat-rate distributor is more efficient at high volume; a commission distributor is more efficient at low volume because the up-front cost is a one-off. An artist generating US$10,000 a year in streaming royalties pays DistroKid US$24.99 (a 0.25% effective rate) and pays CD Baby US$900 (9%). At US$1,000 a year the same artist pays DistroKid US$24.99 (2.5%) and pays CD Baby US$90 (also 9% but a smaller absolute number). The first level of analysis is whether the artist’s expected catalog earnings justify a flat fee or not.

The second level is what the distributor adds beyond moving files to platforms. AWAL’s 15% buys access to a sync team, a marketing infrastructure, a catalog-management portal, and label-services support that an independent artist working alone would need to assemble themselves. Symphonic’s higher-touch tiers do similar. Pure self-distribution through DistroKid is cheap because it is a thin layer; the artist absorbs the work the distributor would otherwise be doing.

The third level is what the distributor does with the revenue once it arrives. This is where a meaningful share of complaint about distribution surfaces. Distributors collect from DSPs in the artist’s name, hold the money in suspense accounts, batch the conversions across multiple platforms and currencies, deduct their fees, and disburse to the artist. The reporting an artist receives can be granular or generic depending on the distributor. The reconciliation between platform statements and distributor statements can be straightforward or impenetrable. Loewenthal’s case against The Orchard, the Sony-owned distributor he served a formal demand letter to over neighbouring rights, sits in this third level. He alleges that since 2017, The Orchard has been collecting Tora’s neighbouring rights without authorisation, and that the distributor admitted in writing in 2022 that it did not have the right to collect, but continued to do so regardless.

A working artist’s actual take-home from a streaming dollar depends on which distribution model they operate under. On a traditional major-label deal, where the label owns the master and the standard 80/20 split runs in the label’s favour, an artist keeps roughly 15 to 25% of master royalties after recoupment. After manager fees (around 15% of what reaches the artist) and lawyer fees on any negotiated deal, an artist on a major-label deal commonly takes home around US$2 to US$3 from a notional US$30 in streaming royalties. On a flat-rate independent distribution setup with no label, no manager, and no lawyer, that same US$30 can leave US$24 or more in the artist’s account after the distributor’s flat fee. Newer artist-friendly deals can land closer to 60/40 or 50/50, and any deal involving recoupable advances comes with the artist’s share withheld until the advance is paid back.

Sync as a counter-example

The streaming side of the business operates on pool calculations and per-stream rates. The sync side, where music is licensed for use in television, film, advertising, and games, operates on individual negotiated fees. The structural difference is worth registering, because it is one of the few corners of the music industry where the artist or their representative knows in advance what a single use of their work will pay.

Indie sync placements in 2025 sit across a wide range. A song in an indie film typically pays US$500 to US$5,000. A placement on a smaller cable or streaming TV show pays US$1,000 to US$10,000. A featured placement on a major streaming-platform original series can pay US$3,000 to US$40,000, depending on the show’s profile and the song’s prominence in the scene. Local TV or radio commercials pay US$1,000 to US$5,000. National US commercials run US$10,000 to US$250,000 or more, with major brand campaigns routinely six figures. Game placements vary widely: a flat buy-out for an indie game might run US$500 to US$15,000; AAA placements or IP-defining tracks command much more. Microsync placements on TikTok, Instagram Reels, or YouTube Shorts pay US$5 to US$500 per use, with low ceilings but high volume.

The sync market has changed shape over the past few years in two ways relevant to indie artists. First, AI-driven catalog matching has lowered the barrier to pitching for placements. Songtradr, Sync Money, and other B2B marketplaces let independent artists upload catalog and have it auto-matched against active briefs from supervisors. Boutique pitchers like Marmoset, Sentric, and Bank Robber Music continue to operate at the higher end with active human supervisors and deeper relationships in the placement community. Second, the major platforms with user-generated content, TikTok and Meta in particular, have moved to blanket-license user uploads of copyrighted music through direct deals with rights holders, paying through DSP-style pools rather than per-placement fees. The “is it sync or is it streaming” question has become genuinely ambiguous on those platforms.

For the working artist, sync remains one of the few music-industry revenue streams where the structural transparency runs in the artist’s favour. A sync agreement is a written contract for a defined use at a specified fee. The artist or publisher can audit it. The check arrives on a stated timetable. None of this is true of streaming pool distributions, and the contrast is not coincidental: sync is where the buyer is a specific party negotiating with a specific rights holder, and the rest of the music industry is increasingly intermediated through pools, samples, and proxies.

The data access problem

The pattern across the chain we have just walked is the same: at every layer, an artist receives a number and is asked to trust it. Distributor statements summarise but do not detail. PRO statements aggregate but do not itemise. DSP dashboards show stream counts but not how those streams were reconciled into a royalty payment. Statements from one collection society do not reconcile cleanly with statements from another. There is no single place an artist can stand and see the full picture of what their work is earning, and what is not reaching them.

This is the structural problem Loewenthal has been documenting.

The mechanism he has been using is privacy law. Most major jurisdictions have, since the late 2010s, established legal rights for individuals to access the personal data that companies hold about them. The European Union’s GDPR Article 15 is the most well-known. Canada’s PIPEDA has a comparable mechanism. The Australian Privacy Act has access provisions in its Australian Privacy Principles. The California Consumer Privacy Act has its own. These laws were not designed for music royalties. They were designed for data privacy in general, which is why they cover any personal information a company holds about an identifiable individual, including, as Loewenthal has been demonstrating, royalty calculations and inter-society correspondence about specific artists.

What he has been able to extract through these mechanisms is significant. From SOCAN, the Canadian PRO, he obtained internal email correspondence between SOCAN and APRA AMCOS discussing how to respond to his data requests. The emails, in his summary, showed APRA telling SOCAN to redirect him back to APRA, while APRA was simultaneously telling him there was nothing more to give. SOCAN’s internal position, in writing, was that it would only release his data if the Canadian Privacy Commissioner legally compelled it. The picture the documents draw is of two collecting societies coordinating, across two countries, on how to respond to one artist’s request to access his own data.

From SESAC, a privately owned American performance rights organisation, he obtained, after a 53-day delay, a formal complaint to the US Federal Trade Commission, and a parallel complaint to the Bavarian data protection authority, a final disclosure file. SESAC participates in the EU-US Data Privacy Framework, which voluntarily commits it to GDPR-equivalent data rights for European artists collecting US royalties. Loewenthal’s filings invoked that framework. The disclosure SESAC eventually delivered in February 2026 contained nearly 11,000 rows of seven years of transactions across Spotify, Apple Music, Pandora, YouTube, Tidal, and others. The file did not contain a single per-stream play count. The only usage metric anywhere in the file was a column called “Num Of Stations,” an aggregate proxy that cannot be used to verify a royalty calculation. SESAC’s Senior Vice President and Counsel stated in writing that this file represents all data in SESAC’s possession concerning Loewenthal. The same correspondence simultaneously stated that SESAC had withheld performance and usage data on privacy grounds on behalf of broadcasters. There appears to be a contradiction here.

The same disclosure surfaced an additional set of problems. A pay hold had been placed on Loewenthal’s SESAC account in February 2019 and never communicated to him, and six years of transactions had been processed against that hold. Concentrated in a single year were 148 unexplained negative royalty entries with no justification provided. SESAC’s final written position, dated February 2026, stated the company had “satisfied any obligations we may have,” wording that leaves more room for ambiguity than a confident compliance position would need.

From APRA AMCOS, he obtained the data set that has become the centrepiece of his public campaign. Submitted as a forensic report to the society itself, it documented 111,977 exact duplicate rows in the royalty data they had supplied: identical play counts for the same song appearing simultaneously across multiple streaming platforms in the underlying records, a configuration he described as statistically impossible. The forensic report also documented hundreds of unexplained negative royalty entries with no justification provided. APRA AMCOS’s response, in his account, was: “this is the best we can do.”

The point that emerges from all of this is not specific to Loewenthal. The point is that the data integrity problems he has documented are not anomalies. They are what an artist sees when they manage to pry the information loose. The reason most artists do not see this is that most artists have not yet filed the relevant privacy access requests. Loewenthal’s “ARTICLE 15” campaign, named after the GDPR provision, is an attempt to encourage them to do so, on the theory that the same coordinating societies will struggle to obscure the same patterns at scale.

The reform he advocates, what he calls “Direct Programmable Licensing,” is technical and concrete. He argues that the industry should move from blanket licensing, which he describes as a system designed for sheet-music distribution in the early 1900s, to a real-time, per-use, transparent payment system enabled by current technology. Whether that specific proposal is the right answer is, as ever, a separate question from whether the current system is the wrong one.

Where the regulators are now

The structural problems documented in the previous sections are not new. What is changing, in 2025 and 2026, is the willingness of regulators in several jurisdictions to act on them.

The most current Australian development is the Australian Competition and Consumer Commission’s draft determination on APRA AMCOS, issued on 12 December 2025. The ACCC is the regulator with statutory authority over APRA’s blanket-licensing arrangement, which requires periodic reauthorisation. The draft determination proposed a five-year reauthorisation, but with substantially stronger transparency conditions than previous rounds. Under the proposed conditions, APRA would be required to undertake periodic consultative reviews of its licence schemes and distribution arrangements, publish more detailed information about how royalties are distributed to members, produce an annual transparency report disclosing rights revenue, operating costs, and payments to members, and continue the Resolution Pathways alternative dispute resolution scheme established under previous conditions. If APRA’s published rate-calculation information is not sufficiently clear, the ACCC reserves the right to require an independent report. Submissions on the draft closed on 16 February 2026, with a final determination expected later in the year. Loewenthal’s supplementary submission to the proceeding, drawing on his GDPR and PIPEDA disclosures, is part of that regulatory record. An internal ACCC email inadvertently received by him reportedly described the coordination between collecting societies as “quite possibly fraud and/or incompetence”; the case officer also noted that this assessment did not change the overall picture of the proceeding. The final determination, when issued, will be a genuine test of how seriously the regulator is willing to act on the structural concerns it has acknowledged in its own internal correspondence.

In the United Kingdom, the Competition and Markets Authority concluded its market study into music streaming in late 2022, finding that artist concerns were not driven by market concentration and declining to make a market investigation reference. Government work on music streaming has continued since under the Department for Culture, Media and Sport, including parliamentary inquiries into streaming economics and ongoing engagement with the relevant industry bodies. No new market study or formal regulatory action has been initiated as of 2026.

In the United States, the Music Modernization Act of 2018 established the Mechanical Licensing Collective, the central body responsible for collecting and distributing mechanical royalties from streaming services for compositions in the US. The MLC has distributed more than US$2.5 billion cumulatively as of late 2024, but roughly half of black-box royalties remain undistributed more than three years after launch. Word Collections, a publishing administrator, has estimated that 25% of MLC collections are not currently being distributed at all. The MLC sued Spotify in 2024 over Spotify’s reclassification of its Premium subscription as an “audiobook bundle” for royalty calculation purposes, a change publishers say cut their mechanical royalties. Governance criticism has continued: in both 2023 and 2024, every voting publisher member of the MLC board was also a member of the National Music Publishers’ Association board.

Also in the United States, California’s AB 2602 came into force on 1 January 2025, restricting the use of AI digital replicas of performers’ voices and likenesses in contracts that lack reasonably specific descriptions of intended use. AB 1836 created a parallel cause of action for unauthorised AI digital replicas of deceased performers in audiovisual works and sound recordings. The federal NO FAKES Act, which would create a federal right of publicity covering AI digital replicas, was reintroduced in the Senate in April 2025 and remains under consideration. None of these laws directly addresses royalty transparency, but they do address the specific question of who controls an artist’s voice and likeness in an AI-driven environment, which is a question the streaming industry is actively grappling with.

In the European Union, the Copyright Directive 2019/790 was finally implemented in all 27 member states by September 2024, when Poland became the last to transpose it. The Directive’s Article 17 makes platforms liable for user-uploaded copyrighted content and is the legal foundation for the licensing deals between TikTok, Meta, and the rights-holder bodies. The European Parliament passed a resolution in January 2024 calling for stronger remuneration rules specifically for music streaming under the Directive; implementation of that resolution at member-state level is ongoing.

The pattern across jurisdictions is consistent. The structural complaints artists have been making for years are now being acknowledged by regulators in writing. The question, in 2026, is whether the acknowledgement turns into action.

What this means in practice

This article has run long because the system is layered, and a layered system does not get clearer by being summarised. The core observations are these.

A song generates revenue under multiple categories at every public use, collected by different organisations on behalf of different rights holders. Most artists know two of those organisations exist and are visible to none of the rest.

Streaming pays through pool calculations that distribute platform-wide revenue proportionally to platform-wide stream counts. The rules of those pools, including thresholds, fraud penalties, promotional commissions, and the difference between paid and ad-supported streams, materially affect what reaches the artist and are not visible at the artist’s level.

Performance rights organisations process billions of usage records annually with imperfect matching, and the consequences of that imperfect matching, in the form of redistributed unmatched royalties, flow disproportionately to the largest rights holders.

Distributors are the layer artists have the most control over, but the layer is also where the most recent contractual disputes about what is being collected, by whom, and on whose behalf, have surfaced.

Privacy law, originally designed for data protection in general, has emerged as one of the few mechanisms by which an artist can extract information about the calculations underlying their own royalties. Loewenthal’s campaign and his pending lawsuit against Universal Music Publishing are the most visible current example of that approach.

Regulators in Australia, the European Union, and the United States have, to varying degrees, acknowledged the structural concerns artists have been raising. The ACCC’s draft determination on APRA AMCOS is the most concrete current proceeding directly addressing PRO transparency. Its final determination, expected later in 2026, will signal how seriously the regulator is willing to act on its own findings.

For an artist trying to navigate this in practice, a small number of things matter more than the rest. Register every composition with a PRO and a mechanical-rights body in your home country. Register every recording with a neighbouring rights body. Read the small print on whichever distributor you use, and understand exactly how that distributor handles streaming threshold rules, fraud penalties, and discovery-mode commissions. Audit your statements when you can; file privacy access requests when you cannot get the information you need any other way. Keep paperwork on every co-write, every producer credit, every featured-artist agreement. Most of the disputes that disappear an artist’s income happen because the underlying paperwork was never written down, not because the system itself failed.

This piece has not taken a position on what the system should look like. The Tape’s editorial position, stated up front, is that working artists are entitled to better information about the economy they work inside. We will keep covering this.