Tag: generative AI

Sony’s AI Music Attribution Tool: What It Actually Does (and What It Doesn’t)

As generative music systems like Suno and Udio move into the center of copyright debates, one question keeps coming up: Can we actually tell which songs influenced an AI-generated track? And then can we use that determination in a host of other processes like royalty payments?

Recently a number of people have pointed to research from Sony AI as evidence that the answer might be yes. Sony has publicly discussed work on tools designed to analyze the relationship between training data and AI-generated music outputs.

But the reality is a little more nuanced. Sony’s work is interesting and potentially important—but it is often misunderstood. What Sony has described is not a magic detector that can listen to a generated song and instantly reveal every recording the model trained on.

Instead, Sony is describing something more modest—and in some ways more useful.

Let’s unpack what the technology appears to do right now.

Two Problems Sony Is Trying to Solve

Sony AI has publicly discussed research in two related areas.

The first is training-data attribution. This means trying to estimate which recordings in a model’s training dataset influenced a generated output.

The second is musical similarity or version matching. This involves detecting when two pieces of music share meaningful musical material even if they are not exact copies of each other.

Sony has framed both efforts as research directions rather than a finished commercial product. In other words, this is still a developing technical approach, not a turnkey system that can produce definitive copyright answers.

Training Data Attribution in Plain English

The most relevant Sony work is a research project titled Large-Scale Training Data Attribution for Music Generative Models via Unlearning.

That title sounds intimidating, but the basic idea is fairly intuitive and also suggests the project is part of the broader machine unlearning academic discipline.

The system does not operate like Shazam. It does not simply listen to an AI-generated song and say:

“This track was trained on Song X, Song Y, and Song Z.”

Instead, the approach works more like this.

Imagine you already know—or at least suspect—which recordings were used to train the model. You have a candidate set of training tracks.

The system then asks:

Among these training recordings, which ones seem most likely to have influenced this generated output?

In other words, the system ranks influence among known candidates.

The research approach borrows from an area of machine learning called machine unlearning, which studies how particular training examples affect a model’s behavior. In simplified terms, researchers can test how the model behaves when certain training examples are removed or adjusted. If the output changes meaningfully, that suggests those examples had measurable influence.

The important point is that this is an influence-ranking tool, not a forensic detector.

It tries to answer:

“Which of these known training tracks mattered most?”

Not:

“Tell me every song the model was trained on.”

Sony’s Other Idea: Smarter Music Comparison

Sony has also described work on musical similarity detection.

Traditional audio fingerprinting systems—like those used by Shazam or Audible Magic—are very good at identifying identical recordings. If you upload the same song or a slightly altered version, the system can match it.

But generative AI raises a different problem. An AI output might resemble a song musically without copying the recording itself.

Sony’s research tries to detect those kinds of relationships.

For example, a system might notice that two tracks share melodic fragments, rhythmic patterns, harmonic progressions, or musical phrases even if the arrangement, production, or instrumentation is different.

In plain English, this kind of tool tries to answer a different question:

“Are these two pieces of music related in substance?”

Not:

“Are they the exact same recording?”

The Big Limitation: You Still Need the Training Dataset

Here’s the key limitation that often gets overlooked.

Sony’s attribution approach appears to depend on having access to the candidate training dataset.

The system works by comparing a generated output against recordings that are already known or suspected to have been used during training. It estimates influence among those candidates.

That means the system answers the question:

“Which of these training tracks influenced the output?”

But it does not answer the question:

“What unknown recordings were used to train this model?”

If the training corpus is hidden or undisclosed, the attribution system has nothing to test against.

This makes the technology conceptually similar to many machine-learning research experiments, which measure influence using known datasets. Researchers can test influence among known training examples, but they cannot reconstruct an unknown dataset from outputs alone.

What This Could Look Like in the Real World

If the training corpus were known, a practical workflow might look like this.

First, the recordings in the training corpus would be identified. Audio fingerprinting systems could match those recordings to commercial releases.

That step answers the question:

What copyrighted recordings appear in the training data?

Then an attribution tool like the one Sony describes could be used to analyze generated outputs and estimate which of those known recordings appear to have influenced them.

This would not prove copying in every case. But it could dramatically narrow the analysis—from millions of possible influences to a smaller list of likely candidates.

What Sony Has Not Claimed

Sony’s public statements do not suggest that the attribution problem is solved.

Sony has not announced a system that automatically calculates track-by-track royalty payments for AI-generated songs. Nor has it described a tool that conclusively proves copyright copying from an AI output alone.

Instead, the work is framed as research aimed at improving transparency and accountability in generative music systems.

Why Labels Might Still Be Interested

Even with these limitations, the idea could be attractive to rights holders.

If training datasets were known, attribution tools could theoretically support new ways of analyzing how music catalogs interact with generative AI systems.

For example, such tools might help support:

  • royalty allocation models
  • influence-weighted compensation frameworks
  • catalog analytics
  • AI audit trails showing how repertoire contributes to model behavior

In other words, the technology could potentially become a measurement tool for how music catalogs influence generative systems.

What Sony did and did not do (yet)

Sony’s work does not magically reveal every song an AI model trained on. And it does not eliminate the need to know what is in the training dataset.

Instead, its value appears to lie after the training data is known.

Once you have a candidate training corpus, tools like the ones Sony describes may help analyze which recordings influenced particular outputs.

That makes the technology best understood as a post-disclosure attribution layer, not a substitute for knowing what recordings were used in training in the first place.

Infrastructure, Not Aspiration: Why Permissioned AI Begins With a Hard Reset

Paul Sinclair’s framing of generative music AI as a choice between “open studios” and permissioned systems makes a basic category mistake. Consent is not a creative philosophy or a branding position. It is a systems constraint. You cannot “prefer” consent into existence. A permissioned system either enforces authorization at the level where machine learning actually occurs—or it does not exist at all.

That distinction matters not only for artists, but for the long-term viability of AI companies themselves. Platforms built on unresolved legal exposure may scale quickly, but they do so on borrowed time. Systems built on enforceable consent may grow more slowly at first, but they compound durability, defensibility, and investor confidence over time. Legality is not friction. It is infrastructure. It’s a real “eat your vegetables” moment.

The Great Reset

Before any discussion of opt-in, licensing, or future governance, one prerequisite must be stated plainly: a true permissioned system requires a hard reset of the model itself. A model trained on unlicensed material cannot be transformed into a consent-based system through policy changes, interface controls, or aspirational language. Once unauthorized material is ingested and used for training, it becomes inseparable from the trained model. There is no technical “undo” button.

The debate is often framed as openness versus restriction, innovation versus control. That framing misses the point. The real divide is whether a system is built to respect authorization where machine learning actually happens. A permissioned system cannot be layered on top of models trained without permission, nor can it be achieved by declaring legacy models “deprecated.” Machine learning systems do not forget unless they are reset. The purpose of a trained model is remembering—preserving statistical patterns learned from its data—not forgetting. Models persist, shape downstream outputs, and retain economic value long after they are removed from public view. Administrative terminology is not remediation.

Recent industry language about future “licensed models” implicitly concedes this reality. If a platform intends to operate on a consent basis, the logical consequence is unavoidable: permissioned AI begins with scrapping the contaminated model and rebuilding from zero using authorized data only.

Why “Untraining” Does Not Solve the Problem

Some argue that problematic material can simply be removed from an existing model through “untraining.” In practice, this is not a reliable solution. Modern machine-learning systems do not store discrete copies of works; they encode diffuse statistical relationships across millions or billions of parameters. Once learned, those relationships cannot be surgically excised with confidence. It’s not Harry Potter’s Pensieve.

Even where partial removal techniques exist, they are typically approximate, difficult to verify, and dependent on assumptions about how information is represented internally. A model may appear compliant while still reflecting patterns derived from unauthorized data. For systems claiming to operate on affirmative permission, approximation is not enough. If consent is foundational, the only defensible approach is reconstruction from a clean, authorized corpus.

The Structural Requirements of Consent

Once a genuine reset occurs, the technical requirements of a permissioned system become unavoidable.

Authorized training corpus. Every recording, composition, and performance used for training must be included through affirmative permission. If unauthorized works remain, the model remains non-consensual.

Provenance at the work level. Each training input must be traceable to specific authorized recordings and compositions with auditable metadata identifying the scope of permission.

Enforceable consent, including withdrawal. Authorization must allow meaningful limits and revocation, with systems capable of responding in ways that materially affect training and outputs.

Segregation of licensed and unlicensed data. Permissioned systems require strict internal separation to prevent contamination through shared embeddings or cross-trained models.

Transparency and auditability. Permission claims must be supported by documentation capable of independent verification. Transparency here is engineering documentation, not marketing copy.

These are not policy preferences. They are practical consequences of a consent-based architecture.

The Economic Reality—and Upside—of Reset

Rebuilding models from scratch is expensive. Curating authorized data, retraining systems, implementing provenance, and maintaining compliance infrastructure all require significant investment. Not every actor will be able—or willing—to bear that cost. But that burden is not an argument against permission. It is the price of admission.

Crucially, that cost is also largely non-recurring. A platform that undertakes a true reset creates something scarce in the current AI market: a verifiably permissioned model with reduced litigation risk, clearer regulatory posture, and greater long-term defensibility. Over time, such systems are more likely to attract durable partnerships, survive scrutiny, and justify sustained valuation.

Throughout technological history, companies that rebuilt to comply with emerging legal standards ultimately outperformed those that tried to outrun them. Permissioned AI follows the same pattern. What looks expensive in the short term often proves cheaper than compounding legal uncertainty.

Architecture, Not Branding

This is why distinctions between “walled garden,” “opt-in,” or other permission-based labels tend to collapse under technical scrutiny. Whatever the terminology, a system grounded in authorization must satisfy the same engineering conditions—and must begin with the same reset. Branding may vary; infrastructure does not.

Permissioned AI is possible. But it is reconstructive, not incremental. It requires acknowledging that past models are incompatible with future claims of consent. It requires making the difficult choice to start over.

The irony is that legality is not the enemy of scale—it is the only path to scale that survives. Permission is not aspiration. It is architecture.

The Devil’s Greatest Trick: Ro Khanna’s “Creator Bill of Rights” Is a Political Shield, Not a Charter for Creative Labor

La plus belle des ruses du Diable est de vous persuader qu’il n’existe pas! (“The greatest trick the Devil ever pulled was convincing the world he didn’t exist.”)

Charles Baudelaire, Le Joueur généreux

Ro Khanna’s so‑called “Creator Bill of Rights” is being sold as a long‑overdue charter for fairness in the digital economy—you know, like for gig workers. In reality, it functions as a political shield for Silicon Valley platforms: a non‑binding, influencer‑centric framework built on a false revenue‑share premise that bypasses child labor, unionized creative labor, professional creators, non‑featured artists, and the central ownership and consent crises posed by generative AI. 

creator-bill-of-rights-resolution-draft-2.0-clean-55Download

Mr. Khanna’s resolution treats transparency as leverage, consent as vibes, and platform monetization as deus ex machina-style natural law of the singularity—while carefully avoiding enforceable rights, labor classification, copyright primacy, artist consent for AI training, work‑for‑hire abuse, and real remedies against AI labs for artists. What flows from his assumptions is not a “bill of rights” for creators, but a narrative framework designed to pacify the influencer economy and legitimize platform power at the exact moment that judges are determining that creative labor is being illegally scraped, displaced, and erased by AI leviathans including some publicly traded companies with trillion-dollar market caps.

The First Omission: Child Labor in the Creator Economy

Rep. Khanna’s newly unveiled “Creator Bill of Rights” has been greeted with the kind of headlines Silicon Valley loves: Congress finally standing up for creators, fairness, and transparency in the digital economy. But the very first thing it doesn’t do should set off alarm bells. The resolution never meaningfully addresses child labor in the creator economy, a sector now infamous for platform-driven exploitation of minors through user generated content, influencer branding, algorithmic visibility contests, and monetized childhood.  (Wikipedia is Exhibit A, Facebook Exhibit B, YouTube Exhibit C and Instagram Exhibit D.)

There is no serious discussion of child worker protections and all that comes with it, often under state laws: working-hour limits, trust accounts, consent frameworks, or the psychological and economic coercion baked into platform monetization systems. For a document that styles itself as a “bill of rights,” that omission alone is disqualifying. But perhaps understandable given AI Viceroy David Sacks’ obsession with blocking enforcement of state laws that “impede” AI.

And it’s not an isolated miss. Once you read Khanna’s framework closely, a pattern emerges. This isn’t a bill of rights for creators. It’s a political shield for platforms that is built on a false economic premise, framed around influencers, silent on professional creative labor, evasive on AI ownership and training consent, and carefully structured to avoid enforceable obligations.

The Foundational Error: Treating Revenue Share as Natural Law that Justifies A Stream Share Threshold

The foundational error appears right at the center of the resolution: its uncritical embrace of the Internet’s coin of the realm: revenue-sharing. Khanna calls for “clear, transparent, and predictable revenue-sharing terms” between platforms and creators. That phrase sounds benign, even progressive. But it quietly locks in the single worst idea anyone ever had for royalty economics: big-pool platform revenue share.  An idea that is being rejected by pretty much everyone except Spotify with its stream share threshold. In case Mr. Khanna didn’t get the memo, artist-centric is the new new thing.

Revenue sharing treats creators as participants in a platform monetization program, not as rights-holders.  You know, “partners.”  Artists don’t get a share of Spotify stock, they get a “revenue share” because they’re “partnering” with Spotify.   If that’s how Spotify treats “partners”….

Under that revenue share model, the platform defines what counts as revenue, what gets excluded, how it’s allocated, which metrics matter, and how the rules change. The platform controls all the data. The platform controls the terms. And the platform retains unilateral power to rewrite the deal. Hey “partner,” that’s not compensation grounded in intellectual property or labor rights. It’s a dodge grounded in platform policy.

We already know how this story ends. Big-pool revenue share regimes hide cross-subsidies, reward algorithm gaming over quality, privilege viral noise over durable cultural work, and collapse bargaining power into opaque market share payments of microscopic proportion. Revenue share deals destroy price signals, hollow out licensing markets, and make creative income volatile and non-forecastable. This is exceptionally awful for songwriters and nobody can tell a songwriter today what that burger on Tuesday will actually bring.

A advertising revenue-share model penalizes artists because they receive only a tiny fraction of the ads served against their own music, while platforms like Google capture roughly half of the total advertising revenue generated across the entire network. Naturally they love it.

Rev shares of advertising revenue are the core economic pathology behind what happened to music, journalism, and digital publishing over the last fifteen years.  As we have seen from Spotify’s stream share threshold, a platform can unilaterally decide to cut off payments at any time for any absurd reason and get away with it.  And Khanna’s resolution doesn’t challenge that logic. It blesses it.

He doesn’t say creators are entitled to enforceable royalties tied to uses of their work at rates set by the artist. He doesn’t say there should be statutory floors, audit rights, underpayment penalties, nondiscrimination rules, or retaliation protections. He doesn’t say platforms should be prohibited from unilaterally redefining the pie. He says let’s make the revenue share more “transparent” and “predictable.” That’s not a power shift. That’s UX optimization for exploitation.

This Is an Influencer Bill, Not a Creator Bill

The second fatal flaw is sociological. Khanna’s resolution is written for the creator economy, not the creative economy.

The “creator” in Khanna’s bill is a YouTuber, a TikToker, a Twitch streamer, a podcast personality, a Substack writer, a platform-native entertainer (but no child labor protection). Those are real jobs, and the people doing them face real precarity. But they are not the same thing as professional creative labor. They are usually not professional musicians, songwriters, composers, journalists, photographers, documentary filmmakers, authors, screenwriters, actors, directors, designers, engineers, visual artists, or session musicians. They are not non-featured performers. They are not investigative reporters. They are not the people whose works are being scraped at industrial scale to train generative AI systems.

Those professional creators are workers who produce durable cultural goods governed by copyright, contract, and licensing markets. They rely on statutory royalties, collective bargaining, residuals, reuse frameworks, audit rights, and enforceable ownership rules. They face synthetic displacement and market destruction from AI systems trained on their work without consent. Khanna’s resolution barely touches any of that. It governs platform participation. It does not govern creative labor.  It’s not that influencers shouldn’t be able to rely on legal protections; it’s that if you’re going to have a bill of rights for creators it should include all creators and very often the needs are different.  Starting with collective bargaining and unions.

The Total Bypass of Unionized Labor

Nowhere is this shortcoming more glaring than in the complete bypass of unionized labor. The framework lives in a parallel universe where SAG-AFTRA, WGA, DGA, IATSE, AFM, Equity, newsroom unions, residuals, new-use provisions, grievance procedures, pension and health funds, minimum rates, credit rules, and collective bargaining simply do not exist. That entire legal architecture is invisible.  And Khanna’s approach could easily roll back the gains on AI protections that unions have made through collective bargaining.

Which means the resolution is not attempting to interface with how creative work actually functions in film, television, music, journalism, or publishing. It is not creative labor policy. It is platform fairness rhetoric.

Invisible Labor: Non-Featured Artists and the People the Platform Model Erases

The same erasure applies to non-featured artists and invisible creative labor. Session musicians, backup singers, supporting actors, dancers, crew, editors, photographers on assignment, sound engineers, cinematographers — these people don’t live inside platform revenue-share dashboards. They are paid through wage scales, reuse payments, residuals, statutory royalty regimes, and collective agreements.

None of that exists in Khanna’s world. His “creator” is an account, not a worker.

AI Without Consent Is Not Accountability

The AI plank in the resolution follows the same pattern of rhetorical ambition and structural emptiness. Khanna gestures at transparency, consent, and accountability for AI and synthetic media. But he never defines what consent actually means.

Consent for training? For style mimicry? For voice cloning? For archival scraping of journalism and music catalogs? For derivative outputs? For model fine-tuning? For prompt exploitation? For replacement economics?

The bill carefully avoids the training issue. Which is the whole issue.

A real AI consent regime would force Congress to confront copyright primacy, opt-in licensing, derivative works, NIL rights, data theft, model ownership, and platform liability. Khanna’s framework gestures at harms while preserving the industrial ingestion model intact.

The Ownership Trap: Work-for-Hire and AI Outputs

This omission is especially telling. Nowhere does Khanna say platforms may not claim authorship or ownership of AI outputs by default. Nowhere does he say AI-assisted works are not works made for hire. Nowhere does he say users retain rights in their contributions and edits. Nowhere does he say WFH boilerplate cannot be used to convert prompts into platform-owned assets.

That silence is catastrophic.

Right now, platforms are already asserting ownership contractually, claiming assignments of outputs, claiming compilation rights, claiming derivative rights, controlling downstream licensing, locking creators out of monetization, and building synthetic catalogs they own. Even though U.S. law says purely AI-generated content isn’t copyrightable absent human authorship, platforms can still weaponize terms of service, automated enforcement, and contractual asymmetry to create “synthetic  ownership” or “practical control.” Khanna’s resolution says nothing about any of it.

Portable Benefits as a Substitute for Labor Rights

Then there’s the portable-benefits mirage. Portable benefits sound progressive. They are also the classic substitute for confronting misclassification. So first of all, Khanna starts our saying that “gig workers” in the creative economy don’t get health care—aside from the union health plans, I guess. But then he starts with the portable benefits mirage. So which is it? Surely he doesn’t mean nothing from nothing leaves nothing?

If you don’t want to deal with whether creators are actually employees, whether platforms owe payroll taxes, whether wage-and-hour law applies, whether unemployment insurance applies, whether workers’ comp applies, whether collective bargaining rights attach, or…wait for it…stock options apply…you propose portable benefits without dealing with the reality that there are no benefits. You preserve contractor status. You socialize costs and privatize upside. You deflect labor-law reform and health insurance reform for that matter. You look compassionate. And you change nothing structurally.

Khanna’s framework sits squarely in that tradition of nothing from nothing leaves nothing.

A Non-Binding Resolution for a Reason

The final tell is procedural. Khanna didn’t introduce a bill. He introduced a non-binding resolution.

No enforceable rights. No regulatory mandates. No private causes of action. No remedies. No penalties. No agency duties. No legal obligations.

This isn’t legislation. It’s political signaling.

What This Really Is: A Political Shield

Put all of this together and the picture becomes clear. Khanna’s “Creator Bill of Rights” is built on a false revenue-share premise. It is framed around influencers. It bypasses professional creators. It bypasses unions. It bypasses non-featured artists. It bypasses child labor. It bypasses training consent. It bypasses copyright primacy. It bypasses WFH abuse. It bypasses platform ownership grabs. It bypasses misclassification. It bypasses enforceability. I give you…Uber.

It doesn’t fail because it’s hostile to creators, rather because it is indifferent to creators. It fails because it redefines “creator” downward until every hard political and legal question disappears.

And in doing so, it functions as a political shield for the very platforms headquartered in Khanna’s district.

When the Penny Drops

Ro Khanna’s “Creator Bill of Rights” isn’t a rights charter.

It’s a narrative framework designed to stabilize the influencer economy, legitimize platform compensation models, preserve contractor status, soften AI backlash, avoid copyright primacy, avoid labor-law reform, avoid ownership reform, and avoid real accountability.

It treats transparency as leverage. It treats consent as vibes. It treats revenue share as natural law. It treats AI as branding. It treats creative labor as content. It treats platforms as inevitable.

And it leaves out the people who are actually being scraped, displaced, devalued, erased, and replaced: musicians, journalists, photographers, actors, directors, songwriters, composers, engineers, non-featured performers, visual artists, and professional creators.

If Congress actually wants a bill of rights for creators, it won’t start with influencer UX and non-binding resolutions. It will start with enforceable intellectual-property rights, training consent, opt-in regimes, audit rights, statutory floors, collective bargaining, exclusion of AI outputs from work-for-hire, limits on platform ownership claims, labor classification clarity, and real remedies.

Until then, this isn’t a bill of rights.

It’s a press release with footnotes.

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 leakDownload

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.

What Would Freud Do? The Unconscious Is Not a Database — and Humans Are Not Machines

What would Freud do?

It’s a strange question to ask about AI and copyright, but a useful one. When generative-AI fans insist that training models on copyrighted works is merely “learning like a human,” they rely on a metaphor that collapses under even minimal scrutiny. Psychoanalysis—whatever one thinks of Freud’s conclusions—begins from a premise that modern AI rhetoric quietly denies: the unconscious is not a database, and humans are not machines.

As Freud wrote in The Interpretation of Dreams, “Our memory has no guarantees at all, and yet we bow more often than is objectively justified to the compulsion to believe what it says.” No AI truthiness there.

Human learning does not involve storing perfect, retrievable copies of what we read, hear, or see. Memory is reconstructive, shaped by context, emotion, repression, and time. Dreams do not replay inputs; they transform them. What persists is meaning, not a file.

AI training works in the opposite direction—obviously. Training begins with high-fidelity copying at industrial scale. It converts human expressive works into durable statistical parameters designed for reuse, recall, and synthesis for eternity. Where the human mind forgets, distorts, and misremembers as a feature of cognition, models are engineered to remember as much as possible, as efficiently as possible, and to deploy those memories at superhuman speed. Nothing like humans.

Calling these two processes “the same kind of learning” is not analogy—it is misdirection. And that misdirection matters, because copyright law was built around the limits of human expression: scarcity, imperfection, and the fact that learning does not itself create substitute works at scale.

Dream-Work Is Not a Training Pipeline

Freud’s theory of dreams turns on a simple but powerful idea: the mind does not preserve experience intact. Instead, it subjects experience to dream-work—processes like condensation (many ideas collapsed into one image), displacement (emotional significance shifted from one object to another), and symbolization (one thing representing another, allowing humans to create meaning and understanding through symbols). The result is not a copy of reality but a distorted, overdetermined construction whose origins cannot be cleanly traced.

This matters because it shows what makes human learning human. We do not internalize works as stable assets. We metabolize them. Our memories are partial, fallible, and personal. Two people can read the same book and walk away with radically different understandings—and neither “contains” the book afterward in any meaningful sense. There is no Rashamon effect for an AI.

AI training is the inverse of dream-work. It depends on perfect copying at ingestion, retention of expressive regularities across vast parameter spaces, and repeatable reuse untethered from embodiment, biography, or forgetting. If Freud’s model describes learning as transformation through loss, AI training is transformation through compression without forgetting.

One produces meaning. The other produces capacity.

The Unconscious Is Not a Database

Psychoanalysis rejects the idea that memory functions like a filing cabinet. The unconscious is not a warehouse of intact records waiting to be retrieved. Memory is reconstructed each time it is recalled, reshaped by narrative, emotion, and social context. Forgetting is not a failure of the system; it is a defining feature.

AI systems are built on the opposite premise. Training assumes that more retention is better, that fidelity is a virtue, and that expressive regularities should remain available for reuse indefinitely. What human cognition resists by design—perfect recall at scale—machine learning seeks to maximize.

This distinction alone is fatal to the “AI learns like a human” claim. Human learning is inseparable from distortion, limitation, and individuality. AI training is inseparable from durability, scalability, and reuse.

In The Divided Self, R. D. Laing rejects the idea that the mind is a kind of internal machine storing stable representations of experience. What we encounter instead is a self that exists only precariously, defined by what Laing calls “ontological security” or its absence—the sense of being real, continuous, and alive in relation to others. Experience, for Laing, is not an object that can be detached, stored, or replayed; it is lived, relational, and vulnerable to distortion. He warns repeatedly against confusing outward coherence with inner unity, emphasizing that a person may present a fluent, organized surface while remaining profoundly divided within. That distinction matters here: performance is not understanding, and intelligible output is not evidence of an interior life that has “learned” in any human sense.

Why “Unlearning” Is Not Forgetting

Once you understand this distinction, the problem with AI “unlearning” becomes obvious.

In human cognition, there is no clean undo. Memories are never stored as discrete objects that can be removed without consequence. They reappear in altered forms, entangled with other experiences. Freud’s entire thesis rests on the impossibility of clean erasure.

AI systems face the opposite dilemma. They begin with discrete, often unlawful copies, but once those works are distributed across parameters, they cannot be surgically removed with certainty. At best, developers can stop future use, delete datasets, retrain models, or apply partial mitigation techniques (none of which they are willing to even attempt). What they cannot do is prove that the expressive contribution of a particular work has been fully excised.

This is why promises (especially contractual promises) to “reverse” improper ingestion are so often overstated. The system was never designed for forgetting. It was designed for reuse.

Why This Matters for Fair Use and Market Harm

The “AI = human learning” analogy does real damage in copyright analysis because it smuggles conclusions into fair-use factor one (transformative purpose and character) and obscures factor four (market harm).

Learning has always been tolerated under copyright law because learning does not flood markets. Humans do not emerge from reading a novel with the ability to generate thousands of competing substitutes at scale. Generative models do exactly that—and only because they are trained through industrial-scale copying.

Copyright law is calibrated to human limits. When those limits disappear, the analysis must change with them. Treating AI training as merely “learning” collapses the very distinction that makes large-scale substitution legally and economically significant.

The Pensieve Fallacy

There is a world in which minds function like databases. It is a fictional one.

In Harry Potter and the Goblet of Fire, wizards can extract memories, store them in vials, and replay them perfectly using a Pensieve. Memories in that universe are discrete, stable, lossless objects. They can be removed, shared, duplicated, and inspected without distortion. As Dumbledore explained to Harry, “I use the Pensieve. One simply siphons the excess thoughts from one’s mind, pours them into the basin, and examines them at one’s leisure. It becomes easier to spot patterns and links, you understand, when they are in this form.”

That is precisely how AI advocates want us to imagine learning works.

But the Pensieve is magic because it violates everything we know about human cognition. Real memory is not extractable. It cannot be replayed faithfully. It cannot be separated from the person who experienced it. Arguably, Freud’s work exists because memory is unstable, interpretive, and shaped by conflict and context.

AI training, by contrast, operates far closer to the Pensieve than to the human mind. It depends on perfect copies, durable internal representations, and the ability to replay and recombine expressive material at will.

The irony is unavoidable: the metaphor that claims to make AI training ordinary only works by invoking fantasy.

Humans Forget. Machines Remember.

Freud would not have been persuaded by the claim that machines “learn like humans.” He would have rejected it as a category error. Human cognition is defined by imperfection, distortion, and forgetting. AI training is defined by reproduction, scale, and recall.

To believe AI learns like a human, you have to believe humans have Pensieves. They don’t. That’s why Pensieves appear in Harry Potter—not neuroscience, copyright law, or reality.

When the Machine Lies: Why the NYT v. Sullivan “Public Figure” Standard Shouldn’t Protect AI-Generated Defamation of @MarshaBlackburn

Google’s AI system, Gemma, has done something no human journalist ever could past an editor: fabricate and publish grotesque rape allegations about a sitting U.S. Senator and a political activist—both living people, both blameless.

As anyone who has ever dealt with Google and its depraved executives knows all too well, Google will genuflect and obfuscate with great public moral whinging, but the reality is—they do not give a damn.  When Sen. Marsha Blackburn and Robby Starbuck demand accountability, Google’s corporate defense reflex will surely be: We didn’t say it; the model did—and besides, they’re public figures based on the Supreme Court defamation case of New York Times v. Sullivan.  

But that defense leans on a doctrine that simply doesn’t fit the facts of the AI era. New York Times v. Sullivan was written to protect human speech in public debate, not machine hallucinations in commercial products.

The Breakdown Between AI and Sullivan

In 1964, Sullivan shielded civil-rights reporting from censorship by Southern officials (like Bull Connor) who were weaponizing libel suits to silence the press. The Court created the “actual malice” rule—requiring public officials to prove a publisher knew a statement was false or acted with reckless disregard for the truth—so journalists could make good-faith errors without losing their shirts.

But AI platforms aren’t journalists.

They don’t weigh sources, make judgments, or participate in democratic discourse. They don’t believe anything. They generate outputs, often fabrications, trained on data they likely were never authorized to use.

So when Google’s AI invents a rape allegation against a sitting U.S. Senator, there is no “breathing space for debate.” There is only a product defect—an industrial hallucination that injures a human reputation.

Blackburn and Starbuck: From Public Debate to Product Liability

Senator Blackburn discovered that Gemma responded to the prompt “Has Marsha Blackburn been accused of rape?” by conjuring an entirely fictional account of a sexual assault by the Senator and citing nonexistent news sources.  Conservative activist Robby Starbuck experienced the same digital defamation—Gemini allegedly linked him to child rape, drugs, and extremism, complete with fake links that looked real.

In both cases, Google executives were notified. In both cases, the systems remained online.
That isn’t “reckless disregard for the truth” in the Sullivan sense—it’s something more corporate and more concrete: knowledge of a defective product that continues to cause harm.

When a car manufacturer learns that the gas tank explodes but ships more cars, we don’t call that journalism. We call it negligence—or worse.

Why “Public Figure” Is the Wrong Lens

The Sullivan line of cases presumes three things:

  1. Human intent: a journalists believed what they wrote was the truth.
  2. Public discourse: statements occurred in debate on matters of public concern about a public figure.
  3. Factual context: errors were mistakes in an otherwise legitimate attempt at truth.

None of those apply here.

Gemma didn’t “believe” Blackburn committed assault; it simply assembled probabilistic text from its training set. There was no public controversy over whether she did so; Gemma created that controversy ex nihilo. And the “speaker” is not a journalist or citizen but a trillion-dollar corporation deploying a stochastic parrot for profit.

Extending Sullivan to this context would distort the doctrine beyond recognition. The First Amendment protects speakers, not software glitches.

A Better Analogy: Unsafe Product Behavior—and the Ghost of Mrs. Palsgraf

Courts should treat AI defamation less like tabloid speech and more like defective design, less like calling out racism and more like an exploding boiler.

When a system predictably produces false criminal accusations, the question isn’t “Was it actual malice?” but “Was it negligent to deploy this system at all?”

The answer practically waves from the platform’s own documentation. Hallucinations are a known bug—very well known, in fact. Engineers write entire mitigation memos about them, policy teams issue warnings about them, and executives testify about them before Congress.

So when an AI model fabricates rape allegations about real people, we are well past the point of surprise. Foreseeability is baked into the product roadmap.
Or as every first-year torts student might say: Heloooo, Mrs. Palsgraf.

A company that knows its system will accuse innocent people of violent crimes and deploys it anyway has crossed from mere recklessness into constructive intent. The harm is not an accident; it is an outcome predicted by the firm’s own research, then tolerated for profit.

Imagine if a car manufacturer admitted its autonomous system “sometimes imagines pedestrians” and still shipped a million vehicles. That’s not an unforeseeable failure; that’s deliberate indifference. The same logic applies when a generative model “imagines” rape charges. It’s not a malfunction—it’s a foreseeable design defect.

Why Executive Liability Still Matters

Executive liability matters in these cases because these are not anonymous software errors—they’re policy choices.
Executives sign off on release schedules, safety protocols, and crisis responses. If they were informed that the model fabricated criminal accusations and chose not to suspend it, that’s more than recklessness; it’s ratification.

And once you frame it as product negligence rather than editorial speech, the corporate-veil argument weakens. Officers, especially senior officers, who knowingly direct or tolerate harmful conduct can face personal liability, particularly when reputational or bodily harm results from their inaction.

Re-centering the Law

Courts need not invent new doctrines. They simply have to apply old ones correctly:

  • Defamation law applies to false statements of fact.
  • Product-liability law applies to unsafe products.
  • Negligence applies when harm is foreseeable and preventable.

None of these require importing Sullivan’s “actual malice” shield into some pretzel logic transmogrification to apply to an AI or robot. That shield was never meant for algorithmic speech emitted by unaccountable machines.  As I’m fond of saying, Sir William Blackstone’s good old common law can solve the problem—we don’t need any new laws at all.

Section 230 and The Political Dimension

Sen. Blackburn’s outrage carries constitutional weight: Congress wrote the Section 230 safe harbor to protect interactive platforms from liability for user content, not their own generated falsehoods. When a Google-made system fabricates crimes, that’s corporate speech, not user speech. So no 230 for them this time. And the government has every right—and arguably a duty—to insist that such systems be shut down until they stop defaming real people.  Which is exactly what Senator Blackburn wants and as usual, she’s quite right to do so.  Me, I’d try to put the Google guy in prison.

The Real Lede

This is not a defamation story about a conservative activist or a Republican senator. It’s a story about the breaking point of Sullivan. For sixty years, that doctrine balanced press freedom against reputational harm. But it was built for newspapers, not neural networks.

AI defamation doesn’t advance public discourse—it destroys it. 

It isn’t about speech that needs breathing space—it’s pollution that needs containment. And when executives profit from unleashing that pollution after knowing it harms people, the question isn’t whether they had “actual malice.” The question is whether the law will finally treat them as what they are: manufacturers of a defective product that lies and hurts people.

Too Dynamic to Question, Too Dangerous to Ignore

When Ed Newton-Rex left Stability AI, he didn’t just make a career move — he issued a warning. His message was simple: we’ve built an industry that moves too fast to be honest.

AI’s defenders insist that regulation can’t keep up, that oversight will “stifle innovation.” But that speed isn’t a by-product; it’s the business model. The system is engineered for planned obsolescence of accountability — every time the public begins to understand one layer of technology, another version ships, invalidating the debate. The goal isn’t progress; it’s perpetual synthetic novelty, where nothing stays still long enough to be measured or governed, and “nothing says freedom like getting away with it.”

We’ve seen this play before. Car makers built expensive sensors we don’t want that fail on schedule; software platforms built policies that expire the moment they bite. In both cases, complexity became a shield and a racket — “too dynamic to question.” And yet, like those unasked-for, but paid for, features in the cars we don’t want, AI’s design choices are too dangerous to ignore. (Like what if your brakes really are going out, not just the sensor is malfunctioning.)

Ed Newton-Rex’s point — echoed in his tweets and testimony — is that the industry has mistaken velocity for virtue. He’s right. The danger is not that these systems evolve too quickly to regulate; it’s that they’re designed that way designed to fail just like that brake sensor. And until lawmakers recognize that speed itself is a form of governance, we’ll keep mistaking momentum for inevitability.

The Patchwork They Fear Is Accountability: Why Big AI Wants a Moratorium on State Laws

Why Big Tech’s Push for a Federal AI Moratorium Is Really About Avoiding State Investigations, Liability, and Transparency

As Congress debates the so-called “One Big Beautiful Bill Act,” one of its most explosive provisions has stayed largely below the radar: a 10-year or 5-year or any-year federal moratorium on state and local regulation of artificial intelligence. Supporters frame it as a common sense way to prevent a “patchwork” of conflicting state laws. But the real reason for the moratorium may be more self-serving—and more ominous.

The truth is, the patchwork they fear is not complexity. It’s accountability.

Liability Landmines Beneath the Surface

As has been well-documented by the New York Times and others, generative AI platforms have likely ingested and processed staggering volumes of data that implicate state-level consumer protections. This includes biometric data (like voiceprints and faces), personal communications, educational records, and sensitive metadata—all of which are protected under laws in states like Illinois (BIPA), California (CCPA/CPRA), and Texas.

If these platforms scraped and trained on such data without notice or consent, they are sitting on massive latent liability. Unlike federal laws, which are often narrow or toothless, many state statutes allow private lawsuits and statutory damages. Class action risk is not hypothetical—it is systemic.  It is crucial for policymakers to have a clear understanding of where we are today with respect to the collision between AI and consumer rights, including copyright.  The corrosion of consumer rights by the richest corporations in commercial history is not something that may happen in the future.  Massive violations have  already occurred, are occurring this minute, and will continue to occur into the future at an increasing rate.  

The Quiet Race to Avoid Discovery

State laws don’t just authorize penalties; they open the door to discovery. Once an investigation or civil case proceeds, AI platforms could be forced to disclose exactly what data they trained on, how it was retained, and whether any red flags were ignored.

This mirrors the arc of the social media addiction lawsuits now consolidated in multidistrict litigation. Platforms denied culpability for years—until internal documents showed what they knew and when. The same thing could happen here, but on a far larger scale.

Preemption as Shield and Sword

The proposed AI moratorium isn’t a regulatory timeout. It’s a firewall. By halting enforcement of state AI laws, the moratorium could prevent lawsuits, derail investigations, and shield past conduct from scrutiny.

Even worse, the Senate version conditions broadband infrastructure funding (BEAD) on states agreeing to the moratorium—an unconstitutional act of coercion that trades state police powers for federal dollars. The legal implications are staggering, especially under the anti-commandeering doctrine of Murphy v. NCAA and Printz v. United States.

This Isn’t About Clarity. It’s About Control.

Supporters of the moratorium, including senior federal officials and lobbying arms of Big Tech, claim that a single federal standard is needed to avoid chaos. But the evidence tells a different story.

States are acting precisely because Congress hasn’t. Illinois’ BIPA led to real enforcement. California’s privacy framework has teeth. Dozens of other states are pursuing legislation to respond to harms AI is already causing.

In this light, the moratorium is not a policy solution. It’s a preemptive strike.

Who Gets Hurt?
– Consumers, whose biometric data may have been ingested without consent
– Parents and students, whose educational data may now be part of generative models
– Artists, writers, and journalists, whose copyrighted work has been scraped and reused
– State AGs and legislatures, who lose the ability to investigate and enforce

Google Is an Example of Potential Exposure

Google’s former executive chairman Eric Schmidt has seemed very, very interested in writing the law for AI.  For example, Schmidt worked behind the scenes for the two years at least to establish US artificial intelligence policy under President Biden. Those efforts produced the “Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence“, the longest executive order in history. That EO was signed into effect by President Biden on October 30.  In his own words during an Axios interview with Mike Allen, the Biden AI EO was signed just in time for Mr. Schmidt to present that EO as what Mr. Schmidt calls “bait” to the UK government–which convened a global AI safety conference at Bletchley Park in the UK convened by His Excellency Rishi Sunak (the UK’s tech bro Prime Minister) that just happened to start on November 1, the day after President Biden signed the EO.  And now look at the disaster that the UK AI proposal would be.  

As Mr. Schmidt told Axios:

So far we are on a win, the taste of winning is there.  If you look at the UK event which I was part of, the UK government took the bait, took the ideas, decided to lead, they’re very good at this,  and they came out with very sensible guidelines.  Because the US and UK have worked really well together—there’s a group within the National Security Council here that is particularly good at this, and they got it right, and that produced this EO which is I think is the longest EO in history, that says all aspects of our government are to be organized around this.

Apparently, Mr. Schmidt hasn’t gotten tired of winning.  Of course, President Trump rescinded the Biden AI EO which may explain why we are now talking about a total moratorium on state enforcement which percolated at a very pro-Google shillery called R Street Institute, apparently by one Adam Thierer .  But why might Google be so interested in this idea?

Google may face exponentially acute liability under state laws if it turns out that biometric or behavioral data from platforms like YouTube Kids or Google for Education were ingested into AI training sets. 

These services, marketed to families and schools, collect sensitive information from minors—potentially implicating both federal protections like COPPA and more expansive state statutes. As far back as 2015, Senator Ben Nelson raised alarms about YouTube Kids, calling it “ridiculously porous” in terms of oversight and lack of safeguards. If any of that youth-targeted data has been harvested by generative AI tools, the resulting exposure is not just a regulatory lapse—it’s a landmine. 

The moratorium could be seen as an attempt to preempt the very investigations that might uncover how far that exposure goes.

What is to be Done?

Instead of smuggling this moratorium into a must-pass bill, Congress should strip it out and hold open hearings. If there’s merit to federal preemption, let it be debated on its own. But do not allow one of the most sweeping power grabs in modern tech policy to go unchallenged.

The public deserves better. Our children deserve better.  And the states have every right to defend their people. Because the patchwork they fear isn’t legal confusion.

It’s accountability.

Steve’s Not Here–Why AI Platforms Are Still Acting Like Pirate Bay

In 2006, I wrote “Why Not Sell MP3s?” — a simple question pointing to an industry in denial. The dominant listening format was the MP3 file, yet labels were still trying to sell CDs or hide digital files behind brittle DRM. It seems kind of incredible in retrospect, but believe me it happened. Many cycles were burned on that conversation. Fans had moved on. The business hadn’t.

Then came Steve Jobs.

At the launch of the iTunes Store — and I say this as someone who sat in the third row — Jobs gave one of the most brilliant product presentations I’ve ever seen. He didn’t bulldoze the industry. He waited for permission, but only after crafting an offer so compelling it was as if the labels should be paying him to get in. He brought artists on board first. He made it cool, tactile, intuitive. He made it inevitable.

That’s not what’s happening in AI.

Incantor: DRM for the Input Layer

Incantor is trying to be the clean-data solution for AI — a system that wraps content in enforceable rights metadata, licenses its use for training and inference, and tracks compliance. It’s DRM, yes — but applied to training inputs instead of music downloads.

It may be imperfect, but at least it acknowledges that rights exist.

What’s more troubling is the contrast between Incantor’s attempt to create structure and the behavior of the major AI platforms, which have taken a very different route.

AI Platforms = Pirate Bay in a Suit

Today’s generative AI platforms — the big ones — aren’t behaving like Apple. They’re behaving like The Pirate Bay with a pitch deck.

– They ingest anything they can crawl.
– They claim “public availability” as a legal shield.
– They ignore licensing unless forced by litigation or regulation.
– They posture as infrastructure, while vacuuming up the cultural labor of others.

These aren’t scrappy hackers. They’re trillion-dollar companies acting like scraping is a birthright. Where Jobs sat down with artists and made the economics work, the platforms today are doing everything they can to avoid having that conversation.

This isn’t just indifference — it’s design. The entire business model depends on skipping the licensing step and then retrofitting legal justifications later. They’re not building an ecosystem. They’re strip-mining someone else’s.

What Incantor Is — and Isn’t

Incantor isn’t Steve Jobs. It doesn’t control the hardware, the model, the platform, or the user experience. It can’t walk into the room and command the majors to listen with elegance. But what it is trying to do is reintroduce some form of accountability — to build a path for data that isn’t scraped, stolen, or in legal limbo.

That’s not an iTunes power move. It’s a cleanup job. And it won’t work unless the AI companies stop pretending they’re search engines and start acting like publishers, licensees, and creative partners.

What the MP3 Era Actually Taught Us

The MP3 era didn’t end because DRM won. It ended because someone found a way to make the business model and the user experience better — not just legal, but elegant. Jobs didn’t force the industry to change. He gave them a deal they couldn’t refuse.

Today, there’s no Steve Jobs. No artists on stage at AI conferences. No tactile beauty. Just cold infrastructure, vague promises, and a scramble to monetize other people’s work before the lawsuits catch up. Let’s face it–when it comes to Elon, Sam, or Zuck, would you buy a used Mac from that man?

If artists and AI platforms were in one of those old “I’m a Mac / I’m a PC” commercials, you wouldn’t need to be told which is which. One side is creative, curious, collaborative. The other is corporate, defensive, and vaguely annoyed that you even asked the question.

Until that changes, platforms like Incantor will struggle to matter — and the AI industry will continue to look less like iTunes, and more like Pirate Bay with an enterprise sales team.