The Sinister Question Spotify Has Not Answered About its AI: What Did They Train On?

In case you missed it, Spotify has apparently been training its own music AI that should allow them to capture some of the AI hype on Wall Street. But it brings back seem bad memories.

There was a time when the music business had a simple rule: “We will never let another MTV build a business on our backs”. That philosophy arose from watching the arbitrage as value created by artists was extracted by platforms that had nothing to do with creating it. That spectacle shaped the industry’s deep reluctance to license digital music in the early years of the internet. “Never” was supposed to mean never.

I took them at their word.

But of course, “never” turned out to be conditional. The industry made exception after exception until the rule dissolved entirely. First came the absurd statutory shortcut of the DMCA safe harbor era. Then YouTube. Then iTunes. Then Spotify. Then Twitter and Facebook, social media. Then TikTok. Each time, platforms were allowed to scale first and renegotiate later (and Twitter still hasn’t paid). Each time, the price of admission for the platform was astonishingly low compared to the value extracted from music and musicians. In many cases, astonishingly low compared to their current market value in businesses that are totally dependent on creatives. (You could probably put Amazon in that category.)

Some of those deals came wrapped in what looked, at the time, like meaningful compensation — headline-grabbing advances and what were described as “equity participation.” In reality, those advances were finite and the equity was often a thin sliver, while the long-term effect was to commoditize artist royalties and shift durable value toward the platforms. That is one reason so many artists came to resent and in many cases openly despise Spotify and the “big pool” model. All the while being told how transformative Spotify’s algorithm is without explaining how the wonderful algorithm misses 80% of the music on the platform.

And now we arrive at the latest collapse of “never”: Spotify’s announcement that it is developing its own music AI and derivative-generation tools.

If you disliked Spotify before, you may loathe what comes next.

This moment is different — but in many ways it is the same fundamental problem MTV created. Artists and labels provided the core asset — their recordings — for free or nearly free, and the platform built a powerful business by packaging that value and selling it back to them. Distribution monetized access to music; AI monetizes the music itself.

According to Music Business Worldwide:

Spotify’s framing appears to offer something of a middle ground. [New CEO] Söderström is not arguing for open distribution of AI derivatives across the internet. Instead, he’s positioning Spotify as the platform where this interaction should happen – where the fans, the royalty pool, and the technology already exist.

Right, our fans and his pathetic “royalty pool.” And this is supposed to make us like you?

The Training Gap

Which brings us to the question Spotify has not answered — the question that matters more than any feature announcement or product demo:

What did they train on?

Was it Epidemic Sound? Was it licensed catalog? Public domain recordings? User uploads? Pirated material?

All are equally possible.

But far more likely to me: Did Spotify train on the recordings licensed for streaming and Spotify’s own platform user data derived from the fans we drove to their service — quietly accumulated, normalized, and ingested into AI over years?

Spotify has not said.

And that silence matters.

The Transparency Gap

Creators currently have no meaningful visibility into whether their work has already been absorbed into Spotify’s generative systems. No disclosure. No audit trail. No licensing registry. No opt-in structure. No compensation framework. The unknowns are not theoretical — they are structural:

Were your recordings used for training?
Do your performances now exist inside model weights?
Was consent ever obtained?
Was compensation ever contemplated?
Can outputs reproduce protected expression derived from your work?

If Spotify trained on catalog licensed to them for an entirely different purpose without explicit, informed permission from rights holders and performers, then AI derivatives are not merely a new feature. They are a massively infringing second layer of value extraction built on top of the first exploitation — the original recordings that creators already struggled to monetize fairly.

This is not innovation. It is recursion.

Platform Data: The Quiet Asset

Spotify possesses one of the largest behavioral and audio datasets in the history of recorded music that was licensed to them for an entirely different purpose — not just recordings, but stems, usage patterns, listener interactions, metadata, and performance analytics. If that corpus was used — formally or informally — as training input for this Spotify AI tool that magically appeared, then Spotify’s AI is built not just on music, but on the accumulated creative labor of millions of artists.

Yet creators were never asked. No notice. No explanation. No disclosure.

It must also be said that there is a related governance question. Daniel Ek’s investment in the defense-AI company Helsing has been widely reported, and Helsing’s systems like all advanced AI depend on large-scale model training, data pipelines, and machine learning infrastructure. Spotify supposedly has separately developed its own AI capabilities.

This raises a narrow but legitimate transparency question: is there any technological, data, personnel, or infrastructure overlap — any “crosstalk” — between AI development connected to Helsing’s automated weapons and the models deployed within Spotify? No public evidence currently suggests such interaction, and the companies operate in different domains, but the absence of disclosure leaves creators and stakeholders unable to assess whether safeguards, firewalls, and governance boundaries exist. Where powerful AI systems coexist under shared leadership influence, transparency about separation is as important as transparency about training itself.

The core issue is not simply licensing. It is transparency. A platform cannot convert custodial access into training rights while declining to explain where its training data came from.

That’s why this quote from MBW belies the usual exceptionally short sighted and moronic pablum from the Spotify executive team:

Asked on the call whether AI music platforms like Suno, Udio and Stability could themselves become DSPs and take share from Spotify, Norström pushed back: “No rightsholder is against our vision. We pretty much have the whole industry behind us.”

Of course, the premise of the question is one I have been wondering about myself—I assume that Suno and Udio fully intend to get into the DSP game. But Spotify’s executive blew right past that thoughtful question and answered a question he wasn’t asked which is very relevant to us: “We have pretty much the whole industry behind us.”

Oh, well, you actually don’t. And it would be very informative to know exactly what makes you say that since you have not disclosed anything about what ever the “it” is that you think the whole industry is behind.

Spotify’s Shadow Library Problem

Across the AI sector, a now-familiar pattern has emerged: Train first. Explain later — if ever.

The music industry has already seen this logic elsewhere: massive ingestion followed by retroactive justification. The question now is whether Spotify — a licensed, mainstream platform for its music service — is replicating that same pattern inside a closed AI ecosystem for which it has no licenses that have been announced.

So the question must be asked clearly:

Is Spotify’s AI derivative engine built entirely on disclosed, authorized training sources? Or is this simply a platform-contained version of shadow-library training?

Because if models ingested:

Unlicensed recordings
User-uploaded infringing material
Catalog works without explicit training disclosure
Performances lacking performer awareness

then AI derivatives risk becoming a backdoor exploitation mechanism operating outside traditional consent structures. A derivative engine built on undisclosed training provenance is not a creator tool. It is a liability gap. You know, kind of like Anna’s Archive.

A Direct Response to Gustav Söderström : What Training Would Actually Be Required?

Launching a true music generation or derivative engine would require massive, structured training, including:

1. Large-Scale Audio Corpus
Millions of full-length recordings across genres, eras, and production styles to teach models musical structure, timbre, arrangement, and performance nuance. Now where might those come from?

2. Stem-Level and Multitrack Data
Separated vocals, instruments, and production layers to allow recombination, remixing, and stylistic transformation.

3. Performance and Voice Modeling
Extensive vocal and instrumental recordings to capture phrasing, tone, articulation, and expressive characteristics — the very elements tied to performer identity.

4. Metadata and Behavioral Signals
Tempo, key, genre, mood, playlist placement, skip rates, and listener engagement data to guide model outputs toward commercially viable patterns.

5. Style and Similarity Encoding
Statistical mapping of musical characteristics enabling the system to generate “in the style of” outputs — the core mechanism behind derivative generation.

6. Iterative Retraining at Scale
Continuous ingestion and refinement using newly available recordings and platform data to improve fidelity and relevance.

7. Funding for all of the above

No generative music system of consequence can be built without enormous training exposure to real recordings and performances, and the expense.

Which returns us to the unresolved question:

Where did Spotify obtain that training data?

Because the issue is not whether Spotify could license training material. The issue is that Spotify has not explained — at all — how its training corpus was assembled.

Opacity is the problem.

Personhood Signals: Training on Recordings Is Training on People

Spotify can describe AI derivatives as “music tools,” but training on recordings is not just training on songs. Recordings contain personhood signals — the distinctive human identifiers embedded in performance and production that let a system learn who someone is (or can sound like), not merely what the composition is.

Personhood signals include (non-exhaustively):

Voice identity markers (timbre, formants, prosody, accent, breath, idiosyncratic phrasing)
Instrumental performance fingerprints (attack, vibrato, timing micro-variance, articulation, swing feel)
Studio-musician signatures (the “nonfeatured” musicians who are often most identifiable to other musicians)
Songwriter styles harmonic signatures, prosodic alignment, and lyric identity markers
Production cues tied to an artist’s brand (adlibs, signature FX chains, cadence habits, recurring delivery patterns)

A modern generative system does not need to “copy Track X” to exploit these signals. It can abstract them — compress them into representations and weights — and then reconstruct outputs that trade on identity while claiming no particular recording was reproduced.

That’s why “licensing” isn’t the real threshold question here. The threshold questions are disclosure and permission:

Did Spotify extract personhood signals from performances on its platform?
Were those signals used to train systems that can output tokenized “sounds like” content?
Are there credible guardrails that prevent the model from generating identity-proximate vocals/instrumental performance?
And can creators verify any of this without having to sue first?

If Spotify’s training data provenance is opaque, then creators cannot know whether their identity-bearing performances were converted into model value which is the beginning of commoditization of music in AI. And when the platform monetizes “derivatives” (aka competing outputs) it risks building a new revenue layer (for Spotify) on top of the very human signals that performers were never asked to contribute.

The Asymmetry Problem

Spotify knows what it trained on. Creators do not. That asymmetry alone is a structural concern.

When a platform possesses complete knowledge of training inputs, model architecture, and monetization pathways — while creators lack even basic disclosure — the bargaining imbalance becomes absolute. Transparency is not optional in this context. It is the minimum condition for legitimacy.

Without it, creators cannot:

Assert rights
Evaluate consent
Measure market displacement
Understand whether their work shaped model behavior
Or even know whether their identity, voice, or performance has already been absorbed into machine systems

As every bully knows, opacity redistributes power.

Derivatives or Displacement?

Spotify frames AI derivatives as creative empowerment — fans remixing, artists expanding, new revenue streams emerging. But the core economic question remains unanswered:

Are these tools supplementing human creation or substituting for it?

If derivative systems can generate stylistically consistent outputs from trained material, then the value captured by the model originates in human recordings — recordings whose role in training remains undisclosed. In that scenario, AI derivatives are not simply tools. They are synthetic competitors built from the creative DNA of the original artists. Kind of like MTV.

The distinction between assistive and substitutional AI is economic, not rhetorical.

The Question That Will Not Go Away

Spotify may continue to speak about AI derivatives in the language of opportunity, scale, and creative democratization. But none of that resolves the underlying issue:

What did they train on?

Until Spotify provides clear, verifiable disclosure about the origin of its training data — not merely licensing claims, but actual transparency — every derivative output carries an unresolved provenance problem. And in the age of generative systems, undisclosed training is a real risk to the artists who feed the beast.

Framed this way, the harm is not merely reproduction of a copyrighted recording; it’s the extraction and commercialization of identity-linked signals from performances potentially impacting featured and nonfeatured performers alike. Spotify’s failure (or refusal) to disclose training provenance becomes part of the harm, because it prevents anyone from assessing consent, compensation, or displacement.

And it makes it impossible to understand what value Spotify wants to license, much less whether we want them to do it at all or train our replacements.

Because maybe, just maybe, we don’t what another Spotify to build a business on our backs.

Anna’s Archive, Spotify, and the Shadow‑Library Playbook: Why Spotify is a Crime Scene

A pirate “preservation” collective best known for shadow libraries now claims to have scraped Spotify at industrial scale, releasing massive metadata dumps and hinting at terabytes of audio to follow. That alone would be alarming. But the deeper significance lies elsewhere: this episode mirrors the same shadow-library acquisition pipeline already at the center of major AI copyright cases like Bartz v. Anthropic and Kadrey v. Meta. It raises uncomfortable questions about Spotify’s security obligations to licensors, the chilling effect of its market power on enforcement, and whether centralized streaming platforms have quietly become the most valuable—and vulnerable—training datasets in the AI economy. Spotify sold itself as the cure for BitTorrent piracy but may have become the back door for the next generation of AI scraping.

Anna’s Archive, a pirate-adjacent “preservation” collective best known for shadow‑library aggregation is now claiming it has scraped Spotify at scale starting before July 2025—releasing a metadata torrent and promising bulk release of audio files measured in the hundreds of terabytes. Spotify has acknowledged unauthorized access, describing scraping of public metadata and illicit tactics used to circumvent DRM to access at least some audio files.

Here’s what Anna’s Blog said:

Spotify leak Download

The immediate story will be framed as piracy. But for Spotify, the real exposure is corporate and contractual. As a public company, Spotify has to manage security and platform-risk disclosures in a way that doesn’t mislead investors, and it has to satisfy rightsholders that its delivery architecture meets the security commitments embedded in licensing practice. If tens of millions of full-length files and user-linked playlist metadata were extractable at scale, the question isn’t only who infringed—it’s whether Spotify’s controls, incident response, and contractual assurances withstand audit, cure, and reputational scrutiny.

This is not just a piracy story or corporate disclosure embarrassment for Spotify. It is a familiar pattern for anyone tracking the book‑dataset litigation side of generative AI: shadow libraries as industrial infrastructure. And that sounds like billion with a B.

THE SHADOW‑LIBRARY PIPELINE: FROM BOOKS TO MUSIC

In the major AI copyright cases involving books like Bartz v. Anthropic and Kadrey v. Meta, the recurring issue has not merely been whether training is fair use, but how the training corpus was acquired in the first place. Allegations in cases such as Bartz v. Anthropic and Kadrey v. Meta focus on bulk ingestion of pirated works from shadow libraries such as LibGen, Z‑Library, and Anna’s Archive as an upstream act distinct from downstream model training and fair use. It’s this ingestion issue that has Anthropic offering a $1.5 billion with a B settlement to authors (which itself is based on the 1999 statutory damages amendment to the Copyright Act that to deter…wait for it…CD ripping).

The Spotify scrape follows the same architectural logic. Swap “books” for “tracks” and the structure seems identical: a centralized, mirrored corpus assembled outside the licensing system, optimized for scale, and perfectly suited for machine‑learning pipelines. Voila.

THE ENTERPRISE TURN: AI ACCESS FOR A FEE

What distinguishes this moment is that shadow libraries are no longer operating solely as donation‑funded activist projects to save humanity. Anna’s Archive has openly described an enterprise‑style access tier, offering high‑speed bulk access to AI labs and other institutional users in exchange for large “donations.” You know, to save humanity. Ahem…

This AI connection reframes shadow libraries as industrial suppliers in my view. The Anna’s rhetoric may be preservation for the good of mankind, but whatever you believe the motives are, the product is dataset procurement—precisely the bottleneck facing AI developers seeking massive, curated, labeled corpora.

SPOTIFY’S SECURITY OBLIGATIONS — AND WHY MARKET POWER MATTERS

Spotify is not just a random consumer app. It is a licensed distribution platform bound by dense contractual obligations to labels, distributors, and other rightsholders. It likely has made substantial commitments to labels, particularly major labels and likely publishers, too regarding its security commitments. Large‑scale scraping raises questions that go well beyond copyright:

• What controls existed to prevent automated extraction at scale?

• What anomaly detection or rate limiting failed?

• What representations were made to licensors and licensees about safeguarding licensed content?

Here, Spotify’s monopoly (or at least dominant) market power becomes its most effective shield. Even if licensors could plausibly argue that security covenants were breached, many are economically dependent on Spotify’s distribution and algorithmic visibility. Enforcement is chilled not by law, but by leverage.

THE DOG THAT DIDN’T BARK: USER DATA

Public reporting has focused on sound recording catalog metadata and audio files. But any incident involving large‑scale access naturally raises a secondary question: what else was accessible through the same pathways? There is no public confirmation that user data has been posted. But “not posted” is not the same as “not taken.”

Listening histories, playlists, device identifiers, or internal engagement metrics would be far more valuable off‑market than in a public torrent.

SPOTIFY’S SHAREHOLDERS SAY WASSUP?

Public reporting that Spotify’s systems were scraped at scale—whether limited to metadata or extending to audio—raises issues that go beyond piracy narratives and into material risk disclosure territory. Spotify is a licensed platform whose core business depends on representations to labels, distributors, and artists about safeguarding catalog integrity and platform security.

Allegations of large-scale scraping, DRM circumvention, or control failures may implicate not only contractual obligations but also cybersecurity risk factors that a reasonable investor could consider material.

Under SEC guidance on cybersecurity disclosures, companies are expected to disclose material risks and incidents, even where investigations are ongoing, if the nature and potential scope could affect operations, relationships, or revenue. The question is not whether piracy exists—it always has—but whether centralized control failures expose Spotify to rightsholder claims, regulatory scrutiny, or renegotiation pressure that could affect future earnings.

So far (and this may change any minute so stay informed), Spotify’s public statement in media outlets is something like this:

“An investigation into unauthorized access identified that a third party scraped public metadata and used illicit tactics to circumvent DRM to access some of the platform’s audio files. We are actively investigating the incident.”

Spotify is circling the wagons and this nonstatement statement tells you everything you need to know: “Investigating” plus “some audio files” is classic damage control. The spin does three things at once: Narrows scope (“some audio files”) without committing to facts; deflects responsibility (“third party,” “illicit tactics”); avoids quantification (no numbers, no dates, no systems).

But—there is no acknowledgment of scale or contradiction. Spotify does not address Anna’s 86 million file / 300 TB claim; or Anna’s claim that files were Spotify-native OGG Vorbis; the presence of Spotify-specific file artifacts, the inclusion of playlist/user-linked data. They don’t even say “the claims are inaccurate.” They just… don’t…engage.

In fact, they never mention Anna’s Archive at all.

“Not necessarily a breach” is a lawyerly non-answer. By emphasizing that this may not represent a breach of internal infrastructure, Spotify leaves open: credential abuse, client-side exploitation, API misuse, session hijacking, DRM circumvention at the delivery layer which seems like it must have happened. In other words: it doesn’t reassure anyone—investors, licensors, or regulators.

Let’s not forget—Anna’s Archive wrote this up in a detailed blog post that is PUBLISHED ON THE INTERNET. People know about it. It’s believable but rebuttable. So rebut it.

That may be fine for one-day PR triage—but it’s not informative enough for me and it may not be for the SEC, either. And I can’t imagine this is what they are telling the labels or publishers.

So while I was posting this, Billboard released a new statement from Spotify:

“Spotify has identified and disabled the nefarious user accounts that engaged in unlawful scraping. We’ve implemented new safeguards for these types of anti-copyright attacks and are actively monitoring for suspicious behavior. Since day one, we have stood with the artist community against piracy, and we are actively working with our industry partners to protect creators and defend their rights.”

Oh, well “new safeguards” solves everything. In the immortal words of Ronnie Scott, and now back to sleep. This public description indicates an account-based circumvention pathway (as opposed to a traditional server intrusion) and reinforces that the incident may involve large-scale extraction through consumer-facing mechanisms. That context heightens the importance of law enforcement inquiry into the scope of data accessed and retained, the representations made by the scraping entity about the nature and purpose of its collection, and the foreseeable downstream uses of redistributed audio and user-linked data.

Spotify’s subsequent statement that it identified and disabled “nefarious user accounts” points to an account-side extraction pathway (perhaps through an account farm) rather than a traditional server breach. While server-side compromises are generally considered more severe, account-based abuse is a well-known and operationally plausible vector for large-scale data extraction when distributed across many automated accounts over time. A corpus on the order of hundreds of terabytes as Anna claims could be exfiltrated in this manner without triggering immediate public disclosure, particularly if activity was geographically distributed and designed to mimic legitimate streaming behavior. The relevant question is therefore not whether such extraction was technically possible, but when it was detected, how it was scoped internally, and whether public representations accurately reflected the magnitude and nature of the activity.

The new statement is still kind of a nondenial denial that could have been written: “Spotify identified and disabled user accounts that unlawfully scraped its platform. Spotify implemented new safeguards against these anti-copyright attacks and is actively monitoring for suspicious behavior.”

But…from an SEC perspective, this new statement makes it harder for Spotify to treat the episode as purely hypothetical risk. They’ve acknowledged an “attack,” they disabled accounts, and they changed safeguards. That’s the kind of “known event” that can support 10-Q risk factor updates (and possibly more, depending on materiality and contractual fallout).

Near silence may be defensible in the very short term (and getting shorter with each passing moment), but prolonged opacity increases the risk that disclosure comes later, under less favorable conditions.

Even if they said something like this, they’d be better off:

Spotify identified unauthorized access in which a third party scraped public metadata and used illicit tactics to circumvent DRM to access some of the platform’s audio files. Spotify is actively investigating the incident.

At least sentient beings could tell what in the world happened.

ENTER THE SECURITIES AND EXCHANGE COMMISSION

Understand that this is just postponing the inevitable in my view. Spotify’s response is notable less for what it says than for what it avoids. By neither confirming nor rebutting the scale and technical specifics described by Anna’s Archive, the company leaves unresolved whether tens of millions of Spotify-delivered audio files were extractable at scale—and whether existing controls matched the assurances Spotify has long made to rightsholders, users, and investors.

This disclosure probably belongs in a SEC Form 10Q, but it could go in an 8-K which will attract a bunch more attention. Form 8-K (Item 1.05 or 8.01) is likely only required if certain thresholds are met, mostly that Spotify deems the incident is deemed material to investors now, and there is a known impact (financial, operational, or contractual), not just investigation.

Given Spotify’s current public posture (“actively investigating,” “some audio files,” “safeguards”), an 8-K would usually be triggered only if they confirm large-scale audio extraction, licensor disputes or termination notices occur, regulatory action is initiated, or remediation costs or litigation exposure become quantifiable.

Right now, Spotify appears to be deliberately staying below the 8-K threshold, and I can’t say as I blame them. There’s no telling how much of Spotify’s share price is attributed (incorrectly in my view) to Spotify selling itself as a tech play, not a music play. The financial press has yet to pick up this story as of this writing, and most of Spotify’s press is in the “how do you manage your awesomeness” BS because the financial press never has understood the music business.

At this stage, the issue fits squarely in an SEC Form 10-Q risk-factor update rather than an 8-K: the company is acknowledging elevated platform and data-extraction risk without conceding a material event that requires immediate disclosure. That could change.

That risk factor might look something like this:

Risks Related to Unauthorized Access, Data Extraction, and Platform Security

We rely on technical controls, contractual restrictions, and digital rights management technologies to protect our platform, our content catalog, and data derived from user and licensor activity. From time to time, third parties may attempt to circumvent these controls through unauthorized access, scraping, or other illicit techniques.

We have identified instances in which third parties have engaged in unauthorized access to certain platform data, including the scraping of public metadata and the circumvention of technological measures designed to protect audio files. While we are actively investigating the scope and impact of such activity and implementing remedial measures, unauthorized access could expose us to reputational harm, regulatory scrutiny, contractual disputes with licensors, litigation, and increased costs associated with security enhancements and enforcement efforts.

In addition, even where extracted data does not include sensitive personal information, large-scale aggregation of platform data—particularly content files or behaviorally derived metrics—may be repurposed by third parties in ways that we cannot control, including for analytics, competitive modeling, or artificial intelligence training. Such downstream uses could adversely affect our relationships with licensors, artists, and other stakeholders, and may give rise to additional legal, regulatory, or commercial risks.

Any failure, or perceived failure, to prevent or respond effectively to unauthorized access or misuse of our platform could materially and adversely affect our business, financial condition, results of operations, or prospects.

THE FINAL IRONY: FROM BITTORRENT TO AI BACK DOOR

Daniel Ek built Spotify in the long shadow of peer‑to‑peer piracy. Spotify’s founding pitch to the music industry was explicit: centralized licensed access would replace BitTorrent as the dominant mode of music consumption. Spotify was sold as an anti‑piracy intervention right along with its fixed prices and shite royalty.

That argument persuaded rightsholders to centralize their catalogs inside a single platform. If you’ve ever sat in a major label marketing meeting, it’s like there are no platforms outside of Spotify.

As of today, Spotify has a huge target on its back. It represents one of the largest curated music corpora ever assembled—the precise object desired by pirate “archivists” and AI developers alike. And the allegation on the table is not marginal leakage, but industrial‑scale scraping that somehow did not get noticed by Spotify. Or so we are expected to believe.

At the same time, Ek has repositioned himself as a major AI and defense‑sector investor, backing companies like Helsing that frame artificial intelligence as strategic infrastructure for fully automated weapons (which are likely a violation of the Geneva Convention as I have discussed with you before). The AI sector’s most valuable input is not code, but data—large, clean, labeled datasets.

The same founder who persuaded the music industry to centralize its catalog inside Spotify as a solution to piracy now presides over a platform alleged to have left a back door open for the next extraction regime: AI scraping at scale.

Piracy did not disappear.

It centralized.

It professionalized.

And now, it has an enterprise pricing tier.

Fasten your seatbelts.

The Digital End-Cap: How Spotify’s Discovery Mode Turned Payola into Personalization

The streaming economy’s most controversial feature revives the old record-store co-op ad model—only now, the shelf space is algorithmic, the payments are disguised as royalty discounts, and the audience has no idea.

From End-Caps to Algorithms: The Disappearing Line Between Marketing and Curation

In the record-store era, everyone in the business knew that end-caps, dump bins, window clings, and in-store listening stations weren’t “organic” discoveries—they were paid placements. Labels bought the best shelf space, sponsored posters, and underwrote the music piped through the store’s speakers because visibility sold records.

Spotify’s Discovery Mode is that same co-op advertising model reborn in code: a system where royalty discounts buy algorithmic shelf space rather than retail real estate. Yet unlike the physical store, today’s paid promotion is hidden behind the language of personalization. Users are told that playlists and AI DJs are “made just for you,” when in fact those recommendations are shaped by the same financial incentives that once determined which CD got the end-cap.

On Spotify, nothing is truly organic; Discovery Mode simply digitizes the old pay-for-placement economy, blending advertising with algorithmic curation while erasing the transparency that once separated marketing from editorial judgment.

Spotify’s Discovery Mode: The “Inverted Payola”

The problem for Spotify is that it has never positioned itself like a retailer. It has always positioned itself as a substitute for radio, and buying radio is a dangerous occupation. That’s called payola.

Spotify’s controversial “Discovery Mode” is a kind of inverted payola which makes it seem like it smells less than it actually does. Remember, artists don’t get paid for broadcast radio airplay in the US so the incentive always was for labels to bribe DJs because that’s the only way that money entered the transaction. (At one point, that could have included publishers, too, back when publishers tried to break artists who recorded their songs.)

What’s different about Spotify is that streaming services do pay for their equivalent of airplay. When Discovery Mode pays less in return for playing certain songs more, that’s essentially the same as getting paid for playing certain songs more. It’s just a more genteel digital transaction in the darkness of ones and zeros instead of the tackier $50 handshake. The discount is every bit as much a “thing of value” as a $50 bill, with the possible exception that it goes to benefit Spotify stockholders and employees unlike the $50 that an old-school DJ probably just put in his pocket in one of those gigantic money rolls. (For games to play on a rainy day, try betting a DJ he has less than $10,000 in his pocket.)

Music Business Worldwide gave Spotify’s side of the story (which is carefully worded flack talk so pay close attention):Spotify rejected the allegations, telling AllHipHop:

“The allegations in this complaint are nonsense. Not only do they misrepresent what Discovery Mode is and how it works, but they are riddled with misunderstandings and inaccuracies.”

The company explained that Discovery Mode affects only Radio, Autoplay and certain Mixes, not flagship playlists like Discover Weekly or the AI DJ that the lawsuit references.Spotify added: “The complaint even gets basic facts wrong: Discovery Mode isn’t used in all algorithmic playlists, or even Discover Weekly or DJ, as it claims.

The Payola Deception Theory

The emerging payola deception theory against Spotify argues that Spotify’s pay-to-play Discovery Mode constitutes a form of covert payola that distorts supposedly neutral playlists and recommendation systems—including Discover Weekly and the AI DJ—even if those specific products do not directly employ the “Discovery Mode” flag.

The key to proving this theory lies in showing how a paid-for boost signal introduced in one part of Spotify’s ecosystem inevitably seeps through the data pipelines and algorithmic models that feed all the others, deceiving users about the neutrality of their listening experience. That does seem to be the value proposition—”You give us cheaper royalties, we give you more of the attention firehose.”

Spotify claims that Discovery Mode affects only Radio, Autoplay, and certain personalized mixes, not flagship products like enterprise playlists or the AI DJ. That defense rests on a narrow, literal interpretation: those surfaces do not read the Discovery Mode switch. Yet under the payola deception theory, this distinction is meaningless because Spotify’s recommendation ecosystem appears to be fully integrated.

Spotify’s own technical publications and product descriptions indicate that multiple personalized surfaces— including Discover Weekly and AI DJ—are built on shared user-interaction data, learned taste profiles, and common recommendation models, rather than each using entirely independent algorithms. It sounds like Spotify is claiming that certain surfaces like Discover Weekly and AI DJ have cabined algorithms and pristine data sets that are not affected by Discovery Mode playlists or the Discovery Mode switch.

While that may be true, it seems like maintaining that separation would be downright hairy if not expensive in terms of compute. It seems far more likely that Spotify run shared models on shared data, and when they say “Discovery Mode isn’t used in X,” they’re only talking about the literal flag—not the downstream effects of the paid boost on global engagement metrics and taste profiles.

How the Bias Spreads: Five Paths of Contamination

So let’s infer that every surface draws on the same underlying datasets, engagement metrics, and collaborative models. Once the paid boost changes user behavior, it alters the entire system’s understanding of what is popular, relevant, or representative of a listener’s taste. The result is systemic contamination: a payola-driven distortion presented to users as organic personalization. The architecture that would make their strong claim true is expensive and unnatural; the architecture that’s cheap and standard almost inevitably lets the paid boost bleed into those “neutral” surfaces in five possible ways.

The first is through popularity metrics. As much as we can tell from the outside, Discovery Mode artificially inflates a track’s exposure in the limited contexts where the switch is activated. Those extra impressions generate more streams, saves, and “likes,” which I suspect feed into Spotify’s master engagement database.

Because stream count, skip rate, and save ratio are very likely global ranking inputs, Discovery Mode’s beneficiaries appear “hotter” across the board. Even if Discover Weekly or the AI DJ ignore the Discovery Mode flag, it’s reasonable to infer that they still rely on those popularity statistics to select and order songs. Otherwise Spotify would need to maintain separate, sanitized algorithms trained only on “clean” engagement data—an implausible and inefficient architecture given Spotify’s likely integrated recommendation system and the economic logic of Discovery Mode itself which I find highly unlikely to be the case. The paid boost thus translates into higher ranking everywhere, not just in Radio or Autoplay. This is the algorithmic equivalent of laundering a bribe through the system—money buys visibility that masquerades as audience preference.

The second potential channel is through user taste profiles. Spotify’s personalization models constantly update a listener’s “taste vector” based on recent listening behavior. If Discovery Mode repeatedly serves a track in Autoplay or Radio, a listener’s history skews toward that song and its stylistic “neighbors”. The algorithm likely then concludes that the listener “likes” similar artists (even if it’s actually Discover Mode serving the track, not user free will. The algorithm likely feeds those likes into Discover Weekly, Daily Mixes, and the AI DJ’s commentary stream. The user thinks the AI is reading their mood; in reality, it is reading a taste profile that was manipulated upstream by a pay-for-placement mechanism. All roads lead to Bieber or Taylor.

A third route is collaborative filtering and embeddings aka “truthiness”. As I understand it, Spotify’s recommendation architecture relies on listening patterns—tracks played in the same sessions or saved to the same playlists become linked in multidimensional “embedding” space. When Discovery Mode injects certain tracks into more sessions, it likely artificially strengthens the connections between those promoted tracks and others around them. The output then seems far more likely to become “fans of Artist A also like Artist B.” That output becomes algorithmically more frequent and hence “truer” or “truthier”, not because listeners chose it freely, but because paid exposure engineered the correlation. Those embeddings are probably global: they shape the recommendations of Discover Weekly, the “Fans also like” carousel, and the candidate pool for the AI DJ. A commercial distortion at the periphery thus is more likely to reshape the supposedly organic map of musical similarity at the core.

Fourth, the DM boost echoes through editorial and social feedback loops. Once Discovery Mode inflates a song’s performance metrics, it begins to look like what passes for a breakout hit these days. Editors scanning dashboards see higher engagement and may playlist the track in prominent editorial contexts. Users might add it to their own playlists, creating external validation. The cumulative effect is that an artificial advantage bought through Discovery Mode converts into what appears to be organic success, further feeding into algorithmic selection for other playlists and AI-driven features. This recursive amplification makes it almost impossible to isolate the paid effect from the “natural” one, which is precisely why disclosure rules exist in traditional payola law. I say “almost impossible” reflexively—I actually think it is in fact impossible, but that’s the kind of thing you can model in a different type of “discovery” being court-ordered discovery.

Finally, there is the shared-model problem. Spotify has publicly acknowledged that the AI DJ is a “narrative layer” built on the same personalization technology that powers its other recommendation surfaces. In practice, this means one massive model (or group of shared embeddings) generates candidate tracks, while a separate module adds voice or context.

If the shared model was trained on Discovery-Mode-skewed data, then even when the DJ module does not read the Discovery flag, it inherits the distortions embedded in those weights. Turning off the switch for the DJ therefore does not remove the influence; it merely hides its provenance. Unlike AI systems designed to dampen feedback bias, Spotify’s Discovery Mode institutionalizes it—bias is the feature, not the bug. You know, garbage in, garbage out.

Proving the Case: Discovery Mode’s Chain of Causation and the Triumph of GIGO

Legally, there’s a strong argument that the deception arises not from the existence of Discovery Mode itself but from how Spotify represents its recommendation products. The company markets Discover Weekly, Release Radar, and AI DJ as personalized to your taste, not as advertising or sponsored content. When a paid-boost mechanism anywhere in the ecosystem alters what those “organic” systems serve, Spotify arguably misleads consumers and rightsholders about the independence of its curation. Under a modernized reading of payola or unfair-deceptive-practice laws, that misrepresentation can amount to a hidden commercial endorsement—precisely the kind of conduct that the Federal Communications Commission’s sponsorship-identification rules (aka payola rules) and the FTC’s endorsement guides were designed to prevent.

In fact, the same disclosure standards that govern influencers on social media should govern algorithmic influencers on streaming platforms. When Spotify accepts a royalty discount in exchange for promoting a track, that arguably constitutes a material connection under the FTC’s Endorsement Guides. Failing to disclose that connection to listeners could transform Discovery Mode from a personalization feature into a deceptive advertisement—modern payola by another name. Why piss off one law enforcement agency when you can have two of them chase you around the rugged rock?

It must also be said that Discovery Mode doesn’t just shortchange artists and mislead listeners; it quietly contaminates the sainted ad product, too. Advertisers think they’re buying access to authentic, personalized listening moments. In reality, they’re often buying attention in a feed where the music itself is being shaped by undisclosed royalty discounts — a form of algorithmic payola that bends not only playlists, but the very audience segments and performance metrics brands are paying for. Advertising agencies don’t like that kind of thing one little bit. We remember what happened when it became apparent that ads were being served to pirate sites by you know who.

Proving the payola deception theory would therefore likely involve demonstrating causation across data layers: that the presence of Discovery Mode modifies engagement statistics, that those metrics propagate into global recommendation features, and that users (and possibly advertisers) were misled to believe those recommendations were purely algorithmic or merit-based. We can infer that the structure of Spotify’s own technology likely makes that chain not only plausible but possibly inevitable.

In an interconnected system where every model learns from the outputs of every other, no paid input stays contained. The moment a single signal is bought, a strong case can be made that the neutrality of the entire recommendation network is compromised—and so is the user’s trust in what it means when Spotify says a song was “picked just for you.”

Spotify’s Paradox: An Algorithm Can Spot Breakout Artists… But Doesn’t Pay Them

Spotify has long claimed its algorithms can spot breakout tracks and rising artists long before traditional gatekeepers notice. From Discover Weekly to Release Radar, the service positions itself as more than a music service — it’s an engine for musical discovery. You know…exposure.

In interviews, press releases, and annual shareholder letters, Daniel Ek the former pirate and future defense contractor and Spotify have boasted that their data can pinpoint the ‘next big thing,’ making Spotify the best place for discovering and nurturing new talent. They raise money on the issue and their market cap is $150.983B today. So they’re sharing the gold with the artists that they extract value from, yes?

The New Threshold Policy

Well, no. Spotify now imposes a threshold: tracks must reach roughly 1,000 streams per year to qualify for recorded music royalties. But understand this–while they don’t pay on those 1,000 streams, they do collect listener data on them and that powers their algorithm. You know, the algorithm they brag on can spot talent, kid. Because those first 1,000 streams are whatchamacallit…free goods. Or is it breakage? Or maybe a kind of packaging deduction?

Under this no-pay-but-track model, Spotify can recognize early listener activity and harness it for its personalized playlists and marketing efforts — yet it doesn’t pay artists for those early plays until the artist meets an arbitrary 1,000 stream threshold. And realize this–an artist could have 100 tracks with 1 million streams each and 100 tracks with less than 1,000, and they won’t get paid on the less streamed tracks. So the pitch here is the track not the artist, yet another contradiction from the Spotify pitch deck.

The Conflict

This creates a troubling tension:

For Spotify: The first few hundred plays are enough for its algorithm to assess the track, recommend it to listeners, and monetize engagement. And brag on it to investors.

For Artists: Those same early plays don’t count toward royalties. The track can gain exposure across the platform, but the artist sees no income until the threshold is crossed.

What It Means

Aside from they’re scumbags? Well, if Spotify can spot a breakout track when it has only a few hundred plays, why can’t it pay for those plays? Its data and recommendations benefit Spotify — obviously, they brag on it. The free goods data makes the service more attractive, bolstering its story to investors, and enriching its user experience — long before it benefits the artist making the records. (To add insult to injury, they do have to pay the songwriters under the benighted compulsory license which they no doubt are going to try to ditch in the next CRB.)

This paradox exposes a deeper tension in the streaming model: platforms excel at extracting value from listener data but adopt a policy to let them grab that data long before it translates into royalties for creators. Talk about data is the new oil.

This paradox shines a light on how platforms monetize listener data early — and benefit from it — while delaying or denying payments for those same streams. What Spotify promotes as ‘discovery’ often operates as monetization for the platform, leaving the artist with nothing until an arbitrary threshold is crossed. The same is likely true of all the platforms that adopt Spotify’s freebie threshold model in what the antitrust lawyers call conscious parallelism, looking at you Amazon and Deezer.

What is to be Done?

If Spotify truly champions discovery, it should champion fair compensation from the first stream. The ability to spot breakout tracks early must be matched by a commitment to pay for them.

It’s time for a more transparent and equitable approach — one that recognizes the value of every stream and every listener, regardless of its place in the self licking ice cream cone.

What the Algocrats Want You to Believe: Weird Al’s Sandwich

There are five key assumptions that support the streamer narrative and we will look at them each in turn. I introduced assumption #1: Streamers are not in the music business, they are in the data business, and assumption #2 : Spotify crying poor. Today we’ll assess assumption #3–streaming royalties are fair no matter what the artists and songwriters say. Like Weird Al.

Assumption 3: The Malthusian Algebra Claims Revenue Share Royalty Pools Are Fair

A corollary of Assumption 2 is that revenue royalty share deals are fair. The way this scam works is that tech companies want to limit their royalty exposure by allocating a chunk of cash that is capped and throwing that bacon over the cage so the little people can fight over it. This produces an implied per-stream rate instead of negotiating an express per stream rate. Why? So they can tell artists–and more importantly regulators, especially antitrust regulators—all the billions they pay “the music business”, whoever that is.

And yet, very few artists or songwriters can live off of streaming royalties, largely because the “big pool” method of hyper-efficient market share distribution that constantly adds mouths to feed is a race to the bottom. The realities of streaming economics should sound familiar to anyone who studied the work of the British economist and demographer Thomas Malthus. Malthus is best known for his theory on population growth in his 1798 book “An Essay on the Principle of Population”. This theory, often referred to as the Malthusian theory (which is why I call the big pool royalty model the “Malthusian algebra”), posits that population growth tends to outstrip food production, leading to inevitable shortages and suffering because, he argued, while food production increases arithmetically, population grows geometrically.

Signally, the big pool model allows the unfettered growth in the denominator while slowing growth in the revenue to increase market valuation based on a growth story. And, of course, the numerator (the productive output of any one artist) is limited by human capacity.

Per-Stream Rate = [Monthly Defined Revenue x (Your Streams ÷ All Streams)]

If the royalty was actually calculated as a fixed penny rate (as is the mechanicals paid by labels on physical and downloads), no artist would be fighting against all other artists for a scrap. In the true Malthus universe, populations increase until they overwhelm the available food supply, which causes humanity’s numbers to be reversed by pandemics, disease, famine, war, or other deadly problems–a Malthusian race to the bottom. Malthus may have inspired Darwin’s theory of natural selection. As a side note, the real attention to abysmal streaming royalties came during the COVID pandemic–which Malthus might have predicted.

Malthus believed that welfare for the poor, inadvertently encouraged marriages and larger families that the poor could not support1. He argued that the only way to break this cycle was to abolish welfare and championed a welfare law revision in 1834 that made conditions in workhouses less appealing than the lowest-paying jobs. You know, “Are there no prisons?” (Not a casual connection to Scrooge in A Christmas Carol.)

Crucially, the difference between Malthusian theory and the reality of streaming is that once artists deliver their tracks, Daniel Ek is indifferent to whether the streaming economics causes them to “die” or retire or actually starve to actual death. He’s already got the tracks and he’ll keep selling them forever like an evil self-licking ice cream cone.

As Daniel Ek told MusicAlly, “There is a narrative fallacy here, combined with the fact that, obviously, some artists that used to do well in the past may not do well in this future landscape, where you can’t record music once every three to four years and think that’s going to be enough.” This is kind of like TikTok bragging about how few children hung themselves in the latest black out challenge compared to the number of all children using the platform. Pretty Malthusian. It’s not a fallacy; it’s all too true.

Crucially, capping the royalty pool allowed Spotify to also cap their subscription rates for a decade or so. Cheap subscriptions drove Spotify’s growth rate and that also droves up their stock price. You’ll never get rich off of streaming royalties, but Daniel Ek got even richer driving up Spotify’s share price. Daniel Ek’s net worth varies inversely to streaming rates–when he gets richer, you get poorer.

Weird Al’s Streaming Sandwich: Using forks and knives to eat their bacon

This race to the bottom is not lost on artists. Al Yankovic, a card-carrying member of the pantheon of music parodists from Tom Leher to Spinal Tap to the Rutles, released a hysterical video about his “Spotify Wrapped” account.

The legend Weird Al criticizing what Spotify pays artists in the video he made FOR Spotify. 🐐🐐🐐 pic.twitter.com/wASQunepBg
— Ray Padgett (@rayfp) November 29, 2023

Al said he’d had 80 million streams and received enough cash from Spotify to buy a $12 sandwich. This was from an artist who made a decades-long career from—parody. Remember that–parody.

Do you think he really meant he actually got $12 for 80 million streams? Or could that have been part of the gallows humor of calling out Spotify Wrapped as a propaganda tool for…Spotify? Poking fun at the massive camouflage around the Malthusian algebra of streaming royalties? Gallows humor, indeed, because a lot of artists and especially songwriters are gradually collapsing as predicted by Malthus.

The services took the bait Al dangled, and they seized upon Al’s video poking fun at how ridiculously low Spotify payments are to make a point about how Al’s sandwich price couldn’t possibly be 80 million streams and if it were, it’s his label’s fault. (Of course, if you ever worked at a label you know that if you haven’t figured out how anything and everything is the label’s fault, you just haven’t thought about it long enough.)

Nothing if not on message, right? Even if by doing so they commit the cardinal sin—don’t try to out-funny a comedian. Or a parodist. Bad, bad idea. (Classic example is Lessig trying to be funny with Stephen Colbert.) Just because Mom laughs at your jokes doesn’t mean you’re funny. And you run the risk of becoming the gag yourself because real comedians will escalate beyond anywhere you’re comfortable.

Weird Al from UHF

It turns out that I have some insight into Al’s thinking and I can tell you he’s a very, very smart guy. The sandwich gag was a joke that highlights the more profound point that streaming sucks. Remember, Al’s the one who turned Dr. Demento tapes into a brand that’s lasted many years. We’ll see if Spotify’s business lasts as long as Weird Al’s career.

I’d suggest that Al was making the point that if you think of everyday goods, like bacon for example, in terms of how many streams you would have to sell in order to buy a pound of bacon, a dozen eggs, a gallon of gasoline, Internet access, or a sandwich in a nice restaurant, you start to understand that the joke really is on us.

What the Algocrats Want You to Believe: Spotify Crying Poor

There are five key assumptions that support the streamer narrative and we will look at them each in turn. I introduced assumption #1: Streamers are not in the music business, they are in the data business. That shouldn’t be a controversial thought. Today we’ll assess assumption #2–streamers like Spotify can’t make a profit.

Assumption #2: Spotify can’t make a profit.

Spotify commonly tells you that they pay 70% of their “revenue” to “the music business” in the “big pool” royalty method. The assumption they want you to make is that they pay billions and if it doesn’t trickle down to artists and songwriters, it’s not their fault.

Remember The Trichordist Streaming Price Bible? If you recall, the abysmal per-stream rates that made headlines were derived by “a mid-sized indie label with an approximately 350+ album catalog now generating over 1.5b streams annually.” Those penny rates were not the artist share, they were derived at the label level. The artist share had to be even worse. And those rates were in 2020–we’ve since had five years of the expansion of the denominator without an offsetting increase in revenues.

Streamers will avoid discussing penny rates like the plague because the rates are just so awful and paupering. They do this by gaslighting–not only artists and songwriters, but also gaslighting regulators. They will tell you that they pay billions “to the music industry” and don’t pay on a per-stream basis so nothing to see here. But they omit the fact that even if they make a lump sum payment to labels or distributors, those labels or distributors have to break down that lump sum to per stream rates in order to account to their artists. So even if the streamers don’t account on a per-stream basis, there is an implied per-stream rate that is simple to derive. Which brings us full-circle to the Streaming Price Bible no matter how they gaslight that supposed 70% revenue share.

And then there’s a remaining 30% because the “revenue” share would have to sum to 100%, right?. That’s true if you assume that the company’s actual revenue is defined the same way as the “revenue” they share with “the music business”. Is it? I think not. I think the actual revenue is higher, and perhaps much higher than the “revenue” as defined in Spotify’s licensing agreements.

Crucially, Spotify’s cash benefits exceed the “revenue” definition on which they account if you don’t ignore the stock market valuation that has made Daniel Ek a billionaire and many Spotify employees into millionaires. Spotify throws off an awful lot of cash for millionaires and billionaires for a company that can’t make a profit.

Good thing that artists and songwriters got compensated for the value their music added to Spotify’s market capitalization and the monetization of all the fans they send to Spotify, right?

Oh yeah. They don’t.

What the Algocrats Want You to Believe

There are five key assumptions that support the streamer narrative and we will look at them each in turn. Today we’ll assess assumption #1–streamers are not in the music business but they want you to believe the opposite.

Assumption 1: Streamers Are In the Music Business

Streamers like Spotify, TikTok and YouTube are not in the music business. They are in the data business. Why? So they can monetize your fans that you drive to them.

These companies make extensive use of algorithms and artificial intelligence in their business, especially to sell targeted advertising. This has a direct impact on your ability to compete with enterprise playlists and fake tracks–or what you might call “decoy footprints”–as identified by Liz Pelly’s exceptional journalism in her new book (did I say it’s on sale now?).

Signally, while Spotify artificially capped its subscription rates for over ten years in order to convince Wall Street of its growth story, the company definitely did not cap its advertising rates which are based on an auction model like YouTube. Like YouTube, Spotify collects emotional data (analyzing a user social media posts), demographics (age, gender, location, geofencing), behavioral data (listening habits, interests), and contextual data (serving ads in relevant moments like breakfast, lunch, dinner). They also use geofencing to target users by regions, cities, postal codes, and even Designated Market Areas (DMAs). My bet is that they can tell if you’re looking at men’s suits in ML Liddy’s (San Angelo or Ft. Worth).

Why the snooping? They do this to monetize your fans. Sometimes they break the law, such as Spotify’s $5.5 million fine by Swedish authorities for violating Europe’s data protection laws.

They’ll also tell you that streamers are all up in introducing fans to new music or what they call “discovery.” The truth is that they could just as easily be introducing you to a new brand of Spam. “Discovery” is just a data application for the thousands of employees of these companies who form the algocracy who make far more money on average than any songwriter or musician does on average. As Maria Schneider anointed the algocracy in her eponymous Pulitzer Prize finalist album, these are the Data Lords. And I gather from Liz Pelly’s book that it’s starting to look like “discovery” is just another form of payola behind the scenes.

It also must be said that these algocrats tend to run together which makes any bright line between the companies harder to define. For example, Spotify has phased out owning data centers and migrated its extensive data operations to the Google Cloud Platform which means Spotify is arguably entirely dependent on Google for a significant part of its data business. Yes, the dominant music streaming platform Spotify collaborates with the adjudicated monopolist Google for its data monetization operations. Not to mention the Meta pixel class action controversy—”It’s believed that Spotify may have installed a tracking tool on its website called the Meta pixel that can be used to gather data about website visitors and share it with Meta. Specifically, [attorneys] suspect that Spotify may have used the Meta pixel to track which videos its users have watched on Spotify.com and send that information to Meta along with each person’s Facebook ID.”

And remember, Spotify doesn’t allow AI training on the music and metadata on its platform.

Right. That’s the good news.

Does it have an index? @LizPelly’s Must-Read Investigation in “Mood Machine” Raises Deep Questions About Spotify’s Financial Integrity

If you don’t know of Liz Pelly, I predict you soon will. I’ve been a fan for years but I really think that her latest work, Mood Machine: The Rise of Spotify and the Costs of the Perfect Playlist, coming in January by One Signal Publishers, an imprint of Atria Books at Simon & Schuster, will be one of those before and after books. Meaning the world you knew before reading the book was radically different than the world you know afterward. It is that insightful. And incriminating.

We are fortunate that Ms. Pelly has allowed Harper’s to excerpt Mood Machine in the current issue. I want to suggest that if you are a musician or care about musicians, or if you are at a record label or music publisher, or even if you are in the business of investing in music, you likely have nothing more important to do today than read this taste of the future.

The essence of what Ms. Pelly has identified is the intentional and abiding manipulation of Spotify’s corporate playlists. She explains what called her to write Mood Machine:

Spotify, the rumor had it, was filling its most popular playlists with stock music attributed to pseudonymous musicians—variously called ghost or fake artists—presumably in an effort to reduce its royalty payouts. Some even speculated that Spotify might be making the tracks itself. At a time when playlists created by the company were becoming crucial sources of revenue for independent artists and labels, this was a troubling allegation.

What you will marvel at is the elaborate means Ms. Pelly has discovered–through dogged reporting worthy of the great deadline artists–that Spotify undertook to deceive users into believing that playlists were organic. And, it must be said, to deceive investors, too. As she tells us:

For years, I referred to the names that would pop up on these playlists simply as “mystery viral artists.” Such artists often had millions of streams on Spotify and pride of place on the company’s own mood-themed playlists, which were compiled by a team of in-house curators. And they often had Spotify’s verified-artist badge. But they were clearly fake. Their “labels” were frequently listed as stock-music companies like Epidemic, and their profiles included generic, possibly AI-generated imagery, often with no artist biographies or links to websites. Google searches came up empty.

You really must read Ms. Pelly’s except in Harper’s for the story…and did I say the book itself is available for preorder now?

All this background manipulation–undisclosed and furtive manipulation by a global network of confederates–was happening while Spotify devoted substantial resources worthy of a state security operation into programming music in its own proprietary playlists. That programmed music not only was trivial and, to be kind, low brow, but also essentially at no cost to Spotify. It’s not just that it was free, it was free in a particular way. In Silicon Valley-speak, Ms. Pelly has discovered how Spotify disaggregated the musician from the value chain.

What she has uncovered has breathtaking implications, particularly with the concomitant rise of artificial intelligence and that assault on creators. The UK Parliament’s House of Commons Digital, Culture, Media & Sport Committee’s Inquiry into the Economics of Music Streaming quoted me as saying “If a highly trained soloist views getting included on a Spotify “Sleep” playlist as a career booster, something is really wrong.” That sentiment clearly resonated with the Committee, but was my feeble attempt at calling government’s attention to then-only-suspected playlist grift that was going on at Spotify. Ms. Pelly’s book is a solid indictment–there’s that word again–of Spotify’s wild-eyed, drooling greed and public deception.

Ms. Pelly’s work raises serious questions about streaming payola and its fellow-travelers in the annals of crime. The last time this happened in the music business was with Fred Dannen’s 1991 book called Hit Men that blew the lid off of radio payola. That book also sent record executives running to unfamiliar places called “book stores” but for a particular reason. They weren’t running to read the book. They already knew the story, sometimes all too well. They were running to see if their name was in the index.

Like the misguided iHeart and Pandora “steering agreements” that nobody ever investigated which preceded mainstream streaming manipulation, it’s worth investigating whether Spotify’s fakery actually rises to the level of a kind of payola or other prosecutable offense. As the noted broadcasting lawyer David Oxenford observed before the rise of Spotify:

The payola statute, 47 USC Section 508, applies to radio stations and their employees, so by its terms it does not apply to Internet radio (at least to the extent that Internet Radio is not transmitted by radio waves – we’ll ignore questions of whether Internet radio transmitted by wi-fi, WiMax or cellular technology might be considered a “radio” service for purposes of this statute). But that does not end the inquiry. Note that neither the prosecutions brought by Eliot Spitzer in New York state a few years ago nor the prosecution of legendary disc jockey Alan Fried in the 1950s were brought under the payola statute. Instead, both were based on state law commercial bribery statutes on the theory that improper payments were being received for a commercial advantage. Such statutes are in no way limited to radio, but can apply to any business. Thus, Internet radio stations would need to be concerned.

Ms. Pelly’s investigative work raises serious questions of its own about the corrosive effects of fake playlists on the music community including musicians and songwriters. She also raises equally serious questions about Spotify’s financial reporting obligations as a public company.

For example, I suspect that if Spotify were found to be using deception to boost certain recordings on its proprietary playlists without disclosing this to the public, it could potentially raise issues under securities laws, including the Sarbanes-Oxley Act (SOX). SOX requires companies to maintain accurate financial records and disclose material information that could affect investors’ decisions.

Deceptive practices that mislead investors about the company’s performance or business practices could be considered a violation of SOX. Additionally, such actions could lead to investigations by regulatory bodies like the Securities and Exchange Commission (SEC) and potential legal consequences.

Imagine that risk factor in Spotify’s next SEC filing? It might read something like this:

Risk Factor: Potential Legal and Regulatory Actions

Spotify is currently under investigation for alleged deceptive practices related to the manipulation of Spotify’s proprietary playlists. If these allegations are substantiated, Spotify could face significant legal and regulatory actions, including fines, penalties, and enforcement actions by regulatory bodies such as the Securities and Exchange Commission (SEC) and the Federal Trade Commission (FTC). Such actions could result in substantial financial liabilities, damage to our reputation, and a loss of user trust, which could adversely affect our business operations and financial performance.

Let’s not do this again, shall we? Did Daniel Ek become a billionaire because of Spotify’s revenue or profit or because of his stock?

Digital Music News reports “Spotify CEO Daniel Ek Is Richer Than Any Musician—Yes, Even Taylor Swift.” Did that just sneak up or is it really Groundhog Day? Maybe it’s groundhog day in Sweden.

Let’s try this again. Remember that artists and labels get paid a revenue share from Spotify. (So do songwriters, but that’s a whole other conversation.) Before you go any farther, getting a share of revenue is for chumps. But what does that mean, this “revenue”. Consider the definition of “gross revenue” that is common in the negotiated version of these deals:

“Gross Revenues” means, with respect to audio and video streams, all gross revenues directly related to the Services, including but not restricted to (i) all revenues attributable to text and/or graphic display, rich media and “in-stream” advertising revenues (i.e., audio, visual or audiovisual advertisements exhibited before, during or after a stream containing any Label Materials) generated from software client interfaces, widgets or properties through which the Services are made available; (ii) all revenues attributable to CPC-, CPM- and CPA-based advertising, e-commerce and “referral fees”/bounties (including non-refundable advances and guarantees, however characterized) generated via the Services, whether structured as a one-time payment or as a recurring revenue share, but specifically excluding e-commerce, “referral fees”/bounties and like revenue generated from sales of permanent audio and video downloads; (iii) all sponsorships sold by Company or its agents; (iv) solely with respect to the Subscription Services, all subscription income; and (v) any share of traffic or tariff charges for delivery of the Services that Company may be able to secure from telecommunications partners, and (vi) all revenues derived from the sale of data related to End Users and their use of the Services [then less a bunch of deductions]”

Now you can just tell that some smart lawyers somewhere sat down to try to think of all the ways that Spotify could earn revenue so they could include those sources in their deal. What did they miss out?

The stock.

In fairness, they didn’t miss out the stock entirely, they just missed it out from the deal that all the artists got paid on. The stock was dealt with in another contract not connected to the main sound recording license and never the twain shall meet.

But what this approach misses entirely is that once you have sold the stock in a stock grant, you’re done being a shareholder. Unless you get another stock grant, which we will assume hasn’t happened.

Leave aside the issue of trading stock for lower royalties, because it’s actually worse that that–it’s trading a one-time stock bump for a lower long term royalty rates set at a price point you have to keep digging out of.

I’m just a country lawyer from Texas and I’m not as smart as the city fellers, but it seems to me that if you knew going in that the big money was in the stock, why wouldn’t you get some measurement of the increase in the net worth of Daniel Ek or some comparable metric as a money factor in the revenue calculation? Getting a one time stock grant isn’t really the same thing. And I say using Ek’s net worth as a bogey only slightly facetiously. That is a little specific, but let’s be honest. It’s Ek’s net worth that really pisses people off, right? And if our Spotify earnings increased in some relationship to his increase in wealth, we’d all probably feel at least less screwed if not actually better about the whole thing.

But even if you didn’t use that metric but knew and acknowledged that the real value was in the stock and the increase in market capitalization due to artists and songwriters, why would you ever allow yourself to get snowed by Spotify’s poor mouthing about they can’t make a profit when it should have been obvious for the last 10-plus years that Spotify didn’t care about making a profit?

The saving grace is, of course, that it’s a damn good thing we’re never going to let another MTV build a business on our backs.

On the Internet, “Partners” Don’t Hear You Scream: Daniel Ek Makes a “Bundle” From the Value He Won’t Share

Here’s a quote for the ages:

MICHAEL BURRY

One of the hallmarks of mania is the rapid rise and complexity
of the rates of fraud. And did you know they’re going up?

The Big Short, screenplay by Charles Randolph and Adam McKay, based on the book by Michael Lewis

I have often said that if screwups were Easter eggs, Daniel Ek would be the Easter bunny, hop hop hopping from one to the next. I realize that is not consistent with his press agent’s pagan iconography, but it sure seems true to many.

The Bunny’s Bundle

This week was no different. Mr. Ek evidently has a “10b5-1 agreement” in place with Spotify, a common technique for insiders, especially founders, who hold at least 10% of the company’s shares to cash out and get the real money through selling their stock. The agreement establishes predetermined trading instructions for company stock (usually a sale and not a buy so not trading the shares) consistent with SEC rules under Section 10b5 of the Securities and Exchange Act of 1934 covering when the insider can sell. Why does this exist? The rule was established in 2000 to protect Silicon Valley insiders from insider trading lawsuits. Yep, you caught it–it’s yet another safe harbor for the special people.

As MusicBusinessWorldWide reported (thank you, Tim), Mr. Ek sold $118.8 million in shares of Spotify at roughly the same time that Spotify was planning to change the way the company paid songwriters on streaming mechanicals by claiming that its recent audiobook offering made it a “bundle” for purposes of the statutory mechanical rate. That would be the same rate that was heavily negotiated in 2021-22 at great expense to all concerned, not to mention torturing the Copyright Royalty Judges. The rates are in effect for five years, but the next negotiation for new rates is coming soon (called Phonorecords V or PR V for short). We’ll get to the royalty bundle but let’s talk about the cash bundle first.

As Tim notes in MBW, Mr. Ek has had a few recent sales under his 10b5-1 agreement: “Across these four transactions (today’s included), Ek has cashed out approximately $340.5 million in Spotify shares since last summer.” Rough justice, but I would place a small wager that Ek has cashed out in personal wealth all or close to all of the money that Spotify has paid to songwriters (through their publishers) for the same period. In this sense, he is no different that the usual disproportionately compensated CEOs at say Google or Raytheon.

Don’t get me wrong, I don’t begrudge Mr. Ek the opportunity to be a billionaire. I don’t at all. But I do begrudge him the opportunity to do it when the government is his “partner” as it is with statutory mechanical royalties, he benefits from various other safe harbors, has had his lobbyists rewrite Section 115 to avoid litigation in a potentially unconstitutional reach back safe harbor, and he hired the lawyer at the Copyright Office who largely wrote the rules that he’s currently bending. Yes, I do begrudge him that stuff.

And here’s the other thing. When Daniel Ek pulls down $340.5 million as a routine matter, I really don’t want to hear any poor mouthing about how Spotify cannot make a profit because of the royalty payments it makes to artists and songwriters. (Or these days, doesn’t make to some artists.) This is, again, why revenue share calculations are just the wrong way to look at the value conferred by featured and nonfeatured artists and songwriters on the Spotify juggernaut. That’s also the point we made in some detail in the paper I co-wrote with Professor Claudio Feijoo for WIPO that came up in Spain, Hungary, France, Uruguay and other countries.

The Malthusian Algebra Strikes Again

It’s not solely Mr. Ek who is the problem child here, it’s partly the fault of industry negotiators who bought into the idea that what was important was getting a share of revenue based on a model that was almost guaranteed to cause royalties to decline over time. This would be getting a share of revenue from someone who purposely suppressed (and effectively subsidized) their subscription pricing for years and years and years. (See Robert Spencer’s Get Big Fast.). If I were a betting man, I would bet that the reason they subsidized the subscription price was to boost the share price by telling a growth story to Wall Street bankers (looking at you, Goldman Sachs) and retail traders because the subsidized subscription price increased subscribers.

Just a guess.

Now about this bundled subscription issue. One of the fundamental points that I think gets missed in the statutory mechanical licensing scheme is the scheme itself. The fact that songwriters have a compulsory license forced on them for one of their primary sources of income is a HUGE concession that songwriters have been asked to agree to since 1909. That’s right–for over 100 years. A decision that seemed reasonable 100 years ago really doesn’t seem reasonable at all today in a networked world. So start there as opposed to streaming platforms are doing us a favor by paying us at all, Daniel Ek saved the music business, and all the other iconography.

Has anyone seen them in the same room at the same time?

The problem that I have with the Spotify move to bundled subscriptions is that it can happen in the middle of a rate period and at least on the surface has the look of a colorable argument to reduce royalty payments. I think if you asked songwriters what they thought the rule was, to the extent they had focused on it at all after being bombarded with self-congratulatory hoorah, they probably thought that the deal wasn’t change rates without renegotiating or at least coming back and asking.

And they wouldn’t be wrong about that, because it is reasonable to ask that any changes get run by your, you know, “partner.” Maybe that’s where it all goes wrong. Because let me suggest and suggest strongly that it is a big mistake to think of these people as your “partner” if by “partner” you mean someone who treats you ethically and politely, reasonably and in good faith like a true fiduciary.

They are not your partner. Stop using that word.

A Compulsory License is a Rent Seeker’s Presidential Suite

But let’s also point out that what is happening with the bundle pricing is a prime example of the brittleness of the compulsory licensing system which is itself like a motel in the desolate and frozen Cyber Pass with a light blinking “Vacancy: Rent Seekers Wanted” surrounded by the bones of empires lost. Unlike the physical mechanical rate which is a fixed penny rate per transaction, the streaming mechanical is a cross between a Rube Goldberg machine and a self-licking ice cream cone.

The Spotify debacle is just the kind of IED that was bound to explode eventually when you have this level of complexity camouflaging traps for the unwary written into law. And the “written into law” part is what makes the compulsory license process so insidious. When the roadside bomb goes off, it doesn’t just hit the uparmored people before the Copyright Royalty Board–it creams everyone.

Helienne Lindvall, David Lowery and Blake Morgan tried to make this point to the Copyright Royalty Judges in Phonorecords IV. They were not confused by the royalty calculations–they understood them all too well. They were worried about fraud hiding in the calculations the same way Michael Burry was worried about fraud in The Big Short. Except there’s no default swaps for songwriters.

Here’s how the Judges responded, you decide if it’s condescending or if the songwriters were prescient knowing what we know now:

While some songwriters or copyright owners may be confused by the royalties or statements of account, the price discriminatory structure and the associated levels of rates in settlement do not appear gratuitous, but rather designed, after negotiations, to establish a structure that may expand the revenues and royalties to the benefit of copyright owners and music services alike, while also protecting copyright owners from potential revenue diminution. This approach and the resulting rate setting formula is not unreasonable. Indeed, when the market itself is complex, it is unsurprising that the regulatory provisions would resemble the complex terms in a commercial agreement negotiated in such a setting.

PR IV Final Rule at 80452 https://app.crb.gov/document/download/27410

It must be said that there never has been a “commercial agreement negotiated in such a setting” that wasn’t constrained by the compulsory license so I’m not sure what that reference even means. But if what the Judges mean is that the compulsory license approximates what would happen in a free market where the songwriters ran free and good men didn’t die like dogs, the compulsory license is nothing like a free market deal. If they are going to allow services to change their business model in midstream but essentially keep their music offering the same while offloading the cost of their audiobook royalties onto songwriters (and probably labels, too, although maybe not) through a discount in the statutory rate, then there should be some downside protection or another bite at the apple.

Unfortunately, there are neither, which almost guarantees another acrimonious, scorched earth lawyer fest in PR V coming soon to a charnel house near you.

Eject, Eject!

This is really disappointing because it was so avoidable if for no other reason. It’s a great time for someone…ahem…to step forward and head off the foreseeable collision on the billable time highway. I actually think the Judges know that the rate calculation is a farce but are dealing with people who have made too much money negotiating it to ever give it up willingly. If they are looking for a way off the theme park ride run by the evil clown, grab my hand on the next pass and I’ll try to pull you out of the centrifugal force. It won’t be easy.

This inevitable dust up means other work will suffer at the CRB. It must be said in fairness that the Judges seem to find it hard enough to get to the work they’ve committed to according to a recent SoundExchange filing in a different case (SDARS III remand from 2020) brought to my attention by Mr. George Johnson.

That’s not uncharitable–I’m merely noting that when dozens of lawyers in Phonorecords proceedings engage in what many of us feel are absurd discovery excesses, you are–frankly–distracting the Judges from doing their job by making them focus on, well, bollocks. We’ll come back to this issue in future, but I think all members of the CRB bar–the dozens and hundreds of those putting children through college at the CRB bar–need to take a breath and realize that judicial resources at the CRB are a zero sum game. This behavior isn’t fair to the Judges and it’s definitely not fair to the real parties in interest–the songwriters.

Tell the Horse to Open Wider

The answer isn’t to get the judges more money, bigger courtroom, craft services and massages, like a financial printer. Some of that would be nice but it doesn’t solve what I think is the real problem. I’d say that the answer is that the participants remember that the main this is that the main thing has to be the main thing. Ultimately, it’s not about us in the phonorecords proceedings, it’s about the songwriters. How are they served?

A compulsory license in stagflationary times is an incredibly valuable gift, and when you not only look the gift horse in the mouth but ask that it open wide so you can check the molars, don’t be surprised if one day it kicks you.

Music Tech Solutions

Chris Castle's Solutions for Music-Tech Entrepreneurs

Category: Digital Services: Spotify

Anna’s Archive, Spotify, and the Shadow‑Library Playbook: Why Spotify is a Crime Scene

THE SHADOW‑LIBRARY PIPELINE: FROM BOOKS TO MUSIC

THE ENTERPRISE TURN: AI ACCESS FOR A FEE

SPOTIFY’S SECURITY OBLIGATIONS — AND WHY MARKET POWER MATTERS

THE DOG THAT DIDN’T BARK: USER DATA

SPOTIFY’S SHAREHOLDERS SAY WASSUP?

ENTER THE SECURITIES AND EXCHANGE COMMISSION

THE FINAL IRONY: FROM BITTORRENT TO AI BACK DOOR

The Digital End-Cap: How Spotify’s Discovery Mode Turned Payola into Personalization

The streaming economy’s most controversial feature revives the old record-store co-op ad model—only now, the shelf space is algorithmic, the payments are disguised as royalty discounts, and the audience has no idea.

From End-Caps to Algorithms: The Disappearing Line Between Marketing and Curation

Spotify’s Discovery Mode: The “Inverted Payola”

The Payola Deception Theory

How the Bias Spreads: Five Paths of Contamination

Proving the Case: Discovery Mode’s Chain of Causation and the Triumph of GIGO

What the Algocrats Want You to Believe: Weird Al’s Sandwich

What the Algocrats Want You to Believe: Spotify Crying Poor

What the Algocrats Want You to Believe

Does it have an index? @LizPelly’s Must-Read Investigation in “Mood Machine” Raises Deep Questions About Spotify’s Financial Integrity

Let’s not do this again, shall we? Did Daniel Ek become a billionaire because of Spotify’s revenue or profit or because of his stock?

On the Internet, “Partners” Don’t Hear You Scream: Daniel Ek Makes a “Bundle” From the Value He Won’t Share

Menu

Chris Castle's Solutions for Music-Tech Entrepreneurs

The Training Gap

The Transparency Gap

Platform Data: The Quiet Asset

Spotify’s Shadow Library Problem

A Direct Response to Gustav Söderström : What Training Would Actually Be Required?

Personhood Signals: Training on Recordings Is Training on People

The Asymmetry Problem

Derivatives or Displacement?

The Question That Will Not Go Away

Share this:

THE SHADOW‑LIBRARY PIPELINE: FROM BOOKS TO MUSIC

THE ENTERPRISE TURN: AI ACCESS FOR A FEE

SPOTIFY’S SECURITY OBLIGATIONS — AND WHY MARKET POWER MATTERS

THE DOG THAT DIDN’T BARK: USER DATA

SPOTIFY’S SHAREHOLDERS SAY WASSUP?

ENTER THE SECURITIES AND EXCHANGE COMMISSION

THE FINAL IRONY: FROM BITTORRENT TO AI BACK DOOR

Share this:

The streaming economy’s most controversial feature revives the old record-store co-op ad model—only now, the shelf space is algorithmic, the payments are disguised as royalty discounts, and the audience has no idea.

From End-Caps to Algorithms: The Disappearing Line Between Marketing and Curation

Spotify’s Discovery Mode: The “Inverted Payola”

The Payola Deception Theory

How the Bias Spreads: Five Paths of Contamination

Proving the Case: Discovery Mode’s Chain of Causation and the Triumph of GIGO

Share this:

The New Threshold Policy

The Conflict

What It Means

What is to be Done?

Share this:

Share this:

Share this:

Share this:

Share this:

Share this:

Share this:

Menu