On April 23, attorneys for Sony Music Entertainment, Universal Music Group, and Warner filed joint stipulations of dismissal in two federal courts.

The Verizon case had been seeking $2.6 billion in damages. The Altice case had potential damages exceeding $1.6 billion. Both cases dismissed with prejudice, meaning they cannot be refiled. Each side bears its own costs and fees.

Combined damages walked away from inside 72 hours: more than $4.2 billion.

The legal reason is straightforward. On March 25, the Supreme Court ruled 9-0 in Cox Communications v. Sony Music that internet service providers cannot be held contributorily liable for their subscribers’ copyright infringement absent specific intent to facilitate it. That ruling closed off the legal theory both lawsuits depended on.

The structural reason is what matters for the agencies and MCNs managing YouTube creators.

What the Cox ruling actually changed

For two decades, the playbook was the same. Major label sues ISP. Argues the ISP knew users were pirating music and didn’t terminate accounts aggressively enough. Wins or settles for hundreds of millions. Repeats with the next ISP.

The 2019 Cox jury verdict was $1 billion. The Grande Communications verdict was $46.7 million. Sony, UMG, Warner, and ABKCO had filed against Verizon in 2024 seeking $2.6 billion, citing more than 500 subscribers who had each received over 100 infringement notices. The Altice complaint listed 6,700 sound recordings and 4,000 compositions allegedly infringed by Altice subscribers between 2020 and 2023.

In Cox v. Sony, all nine Supreme Court justices sided with the ISP. Justice Thomas, writing for the majority, drew the line: contributory liability requires proof that the provider intended its service to be used for infringement, demonstrated either through active inducement or by operating a service incapable of substantial non-infringing uses. Knowing that some subscribers infringe and continuing to serve them isn’t enough.

Justice Sotomayor, joined by Justice Jackson, concurred in the judgment but criticized the majority. Her concern was that the new rule lets ISPs sell connections to known infringers without facing meaningful liability, gutting the incentives Congress wrote into the DMCA’s safe harbor.

The fallout was immediate. The Supreme Court vacated the $46.7 million Grande verdict and remanded for reconsideration. X cited Cox in a court filing two days after the ruling, arguing music publishers’ long-running case against the platform should be dismissed; that case was paused indefinitely on April 3. Meta invoked Cox in its summary judgment filing against Epidemic Sound. RCN cited Cox in pushing back against active label claims. Then on April 23, the labels jointly dismissed Verizon and Altice.

Inside 30 days, more than $5 billion in active label-versus-ISP litigation either evaporated, paused, or got vacated. That’s the most significant retreat the music industry has made from off-platform copyright enforcement in two decades.

The legislative move that’s already underway

Within ten days of the Cox ruling, Senator Thom Tillis (R-NC) and Representative Zoe Lofgren (D-CA) reportedly began merging their separate site-blocking proposals into one bicameral bill. The unified version would require ISPs and large DNS providers (think Cloudflare, Google’s public DNS) to block foreign piracy sites under court order. Tillis isn’t running for reelection. The deadline for getting the bill through Congress runs out in January 2027.

Site-blocking legislation hasn’t had this kind of bipartisan traction since SOPA derailed in 2012. Whether it actually passes is uncertain. Even if it does, it would create a court-ordered blocking mechanism for foreign piracy sites. That’s a different problem than holding ISPs liable for subscriber piracy. It doesn’t restore the playbook the labels just lost.

The labels’ off-platform enforcement budget is going to be looking for a new home for the next twelve to eighteen months minimum, regardless of what Congress does.

Where the budget actually goes

Litigation budgets don’t go back to shareholders when the cases get dismissed. They reallocate to whatever vehicles are still producing recoverable revenue.

The vehicle that has been scaling in parallel for fifteen years is Content ID. YouTube processed 2.2 billion Content ID claims in 2024 alone. The system has paid out $12 billion to rightsholders since launch. It doesn’t require a verdict, a settlement, or a deposition. A claim placed on a video starts capturing revenue immediately, and the burden of disputing it falls on the creator.

If you’re a label rights operations team and the litigation pipeline you’ve been building toward for a decade just lost its target, the rational move is to expand the mechanism that’s already working. That means more aggressive reference filing for catalog you weren’t actively monetizing before. More claim volume against marginal cases that previously weren’t worth the operational cost. Tighter enforcement of catalog you’d been deprioritizing for years.

This won’t happen with an announcement. There is no press release titled “Universal Pivots Enforcement Strategy.” It happens quietly, through procurement. A label rights ops team gets a budget reallocation. They expand their catalog reference work with a third-party service, or use their internal CMS more aggressively. Six months later, claim volume on creator videos has gone up 15-25%, and nobody has connected it to the Cox ruling.

Roster operators are the ones who’ll feel that shift first, in claim volume, before anyone connects the dots back to a Supreme Court decision six months prior.

What this looks like at agency scale

If you run an agency managing 30 channels averaging three uploads per week, you’re processing roughly 90 uploads per week. Even at a conservative 10% Content ID match rate, that’s 9 new claim decisions per week, each running on its own 5-day clock for revenue recovery.

Now apply a 15-20% increase in claim volume across the back half of 2026. You’re looking at 11-12 new claim decisions weekly instead of 9. That’s another two to three claims per week, or roughly 100 to 150 additional claims over a year.

Most of those will be legitimate. Most won’t materially threaten your operation. The marginal claims, the ones in the gray zone where the answer requires research, are the ones that consume agency time. Each one demands the same workflow. Look up the claimant. Check the licensing chain. Decide whether to dispute. File before the 5-day window closes. Track the response. Each cycle takes hours of operational time.

Multiply that by 100-150 incremental cases over a year and you’re not talking about a copyright problem. You’re talking about a staffing problem you didn’t budget for.

The agencies I talk to are mostly running this manually. Someone on ops handles claims as they come in. There is no internal claimant database. No pattern recognition across the roster. No way to surface that a particular claimant has filed against three other channels in the network and disputes are succeeding 80% of the time.

That gap has been a problem for a while. The Cox ruling compresses the timeline on when it becomes urgent.

What to do in the next 90 days

Pull twelve weeks of historical Content ID claim data across your roster. Track the weekly claim rate, the median time to resolution, and the concentration by claimant. The point is to establish a baseline now, before any shift starts. If claim volume goes up 18% in October, you’ll only know if you can compare it to April. Without a baseline, that movement is invisible until it’s already painful.

Audit your dispute decision workflow before you change it. When a claim lands on a creator’s video Tuesday at 2pm, who reviews it, who decides whether to dispute, and what’s the maximum elapsed time before a decision is made? If the honest answer is “depends who’s around,” that’s a process gap, and it’ll get expensive when claim volume rises. Document the workflow even if you don’t redesign it. You can’t fix what you haven’t written down.

Build a claimant reference list. The major labels (UMG, Sony, Warner, BMG, Concord) account for the largest legitimate volume. Below them are the rights management companies (HAAWK/Identifyy, AdRev, Audiam, and others) that handle catalog on behalf of multiple rightsholders. Then there is the long tail of independent claimants. When a claim comes in from a name you don’t recognize, the first question should be “have we seen this name before, and how did the disputes resolve.” Without an internal record, every claim is a cold start.

None of this is sophisticated. All of it can be done in a spreadsheet by an ops associate over two weeks. The agencies that do it before claim volume rises will be the ones that handle the next twelve months without scrambling.

This is the work ClearVerse is built to make easier. Claimant histories instead of cold lookups. Legal precedent on similar disputes instead of guesswork on whether to fight. Portfolio-level visibility for agencies running rosters of 20, 50, 100+ channels.

YouTube tells you a claim was filed. ClearVerse tells you what to do about it.

See how it works!

If you run a roster, what’s your weekly Content ID claim volume right now? Is it different from where it was 90 days ago? And do you have a way to actually know if it changes, or are you running on feel? Hit reply. I read all of them.

— Christian

P.S. — The Cox ruling closed one front. The follow-on effects haven’t shown up in your dashboard yet, which is exactly when there’s still time to prepare for them. Forward this to anyone managing YouTube creators who needs to be paying attention to where this goes next.

The 5-day window closed on Sunday. They lost four days of recoverable revenue. Permanently.

This is not fine print. It is a documented mechanic in YouTube’s Content ID system that most creators and agencies either don’t know exists or don’t understand precisely enough to act on in time. And because YouTube videos earn at their highest rate in the first several days after upload, the window always closes at the worst possible moment.

How the mechanic actually works

When Content ID places a claim on your video, YouTube begins holding the revenue your video generates. That held revenue does not sit there indefinitely. What happens to it, and whether you can recover any of it, depends entirely on what you do and when you do it.

YouTube states this plainly in their own Help Center documentation:

If you dispute a claim within five days of when it was placed, YouTube holds all revenue from Day 1 forward while the dispute is pending. Win the dispute, and you get everything back to the original claim date.

If you dispute after five days, YouTube only holds revenue going forward from the date you file. Everything your video earned between Day 1 and the day you finally decided to act is gone. The claimant keeps it, regardless of how the dispute ultimately resolves.

If you do nothing within five days, the held revenue is released directly to the claimant.

Not still processing. Not waiting for review. Released. To the claimant.

There is a second window worth understanding too. Once you file a dispute, the claimant has 30 days to respond. They can release the claim, reinstate it, or escalate to a formal copyright removal request. If they reinstate and you want to appeal, that window tightens to 7 days. For videos that are blocked outright, YouTube does allow creators to skip the initial dispute and go straight to appeal, which gets you a response in 7 days instead of 30. But none of those downstream windows matter if the first 5 days have already passed without a decision.

Why this lands when it hurts the most

Content ID does not wait for a video to find its audience before scanning it. YouTube processes uploads for copyright matches within minutes of a video going live, often before the first subscriber notification has hit anyone’s feed. A claim can appear on a new video before it has earned its first dollar.

For most channels, view velocity is highest in the first 24 to 72 hours after upload. That is when subscriber notifications push content to existing audiences, when the algorithm surfaces it to new viewers, when social sharing builds. After that initial window, views typically plateau unless the video catches a second wave through search or recommendations.

Which means a copyright claim placed on a new video arrives at precisely the moment the video is making the most money, and the 5-day decision window closes right in the middle of that peak earning period.

A creator who misses the window by 48 hours is not just losing two days of revenue. Depending on the video, they may be losing a meaningful percentage of everything that video will ever earn.

The numbers that make this a systemic problem

YouTube processed 2.2 billion Content ID claims in 2024, a new record. Over 99% were placed automatically, without a human reviewing them before they went out.

Less than 0.5% of those claims were disputed.

That number deserves a closer look. It does not mean that 99.5% of claims are valid and creators are correctly accepting them. The evidence points in a different direction. Of the claims that were actually disputed, approximately 60 to 65% resolved in favor of the uploader, according to YouTube’s own transparency reports. The majority of the time, when a creator fights a claim, they win.

What that combination tells you is that most creators are not accepting claims because the claims are correct. They are accepting them because they do not know how to dispute effectively, do not know whether the dispute will hold up, or did not act within the window before the decision calculus changed.

Meanwhile, rightsholders collected on claims that were never challenged. YouTube has paid out $12 billion total to rightsholders through Content ID since the system launched. That $12 billion includes revenue from disputed claims that resolved in the rightsholder’s favor, but it also includes revenue from claims that were never disputed at all, including claims that would have been released had the creator filed on time.

What this looks like at agency scale

For an individual creator, missing the 5-day window on one video is painful but contained. For an agency managing 20 or 30 channels with multiple uploads per week, the math compounds quickly.

Take an agency running 25 channels averaging two uploads each per week. That is 50 uploads per week. If only 10% of those uploads receive a Content ID claim in the first week (a conservative number for channels that regularly use licensed music), the agency is managing five new claim decisions every week, each with its own 5-day clock, each arriving during or immediately after the video’s highest-earning window.

That is not a copyright problem. That is a workflow problem. And most agencies do not have a structured workflow for it. They have someone who handles copyright when it comes up, which means handling it reactively, which means windows sometimes close on weekends, during high-volume upload periods, or simply while everyone is focused on the next piece of content.

Missing the window on a video with strong early performance can mean losing hundreds or thousands of dollars per video that cannot be recovered regardless of how the dispute eventually resolves. Across a full roster over a year, that is a number worth paying attention to.

What to actually do when a claim lands

Treat the claim date as Day 1, not the day you notice it. YouTube timestamps the claim, and that is the clock that matters. Notifications can lag. Go directly to YouTube Studio to check the timestamp rather than relying on email.

Before doing anything else, look up who filed the claim. YouTube shows you the claimant’s name in Studio. Some claimants are major labels or publishers with well-documented catalogs. Others are third-party rights management companies whose catalogs include public domain content they have registered with varying degrees of legitimacy. The claimant’s identity is the first signal about whether a dispute is worth pursuing and how strong your position is.

Understand the claim type before you act. A monetization claim redirects your revenue while the video stays live. A block claim takes the video down in some or all countries. The dispute paths and the risk profiles are different for each. For block claims, YouTube allows creators to escalate directly to appeal, bypassing the initial 30-day dispute period. That shortens resolution to 7 days. But if the appeal is rejected, the claimant can escalate to a formal copyright removal request, which can result in a copyright strike on your channel. Faster is not always safer.

Prepare your documentation before you file, not after. License agreements, purchase receipts, timestamps showing when you acquired the audio, confirmation emails from the platform you licensed through. If you licensed a track from a third-party service, that paper trail is the foundation of your dispute. Filing without it often results in reinstatement, which limits your options going forward.

Understand what fair use actually protects before you invoke it. YouTube cannot evaluate fair use claims; only a court can make that determination. Disputing on fair use grounds when the case is weak can result in reinstatement and leave you with fewer options than if you had approached it differently from the start. Fair use is a real defense, but it requires a real argument.

The pattern I hear consistently from agencies and creators who have dealt with this is the same. A claim comes in. The decision of what to do feels uncertain. Someone puts it on the list to handle. Two days pass. Three days. By the time there is clarity on whether the claim is legitimate and what the right move is, the window has narrowed or closed.

YouTube flags the claim. The 5-day clock starts immediately. Whether you recover the revenue from those first critical days depends entirely on a decision that has to be made before most people even fully understand the situation.

That is the gap. Not the dispute process itself, which is navigable once you know it. The gap is the time between “a claim just appeared on my video” and “I know what to do about it and I am doing it.”

If you are managing a creator roster, I would like to know how you handle this right now. When a claim lands, is there a process in place, or does it get handled case by case? And how often does a claim arrive in the first 48 hours after a video goes live?

Hit reply. I read all of them.

— Christian

P.S. ClearVerse is building the decision layer for exactly this moment, helping agencies understand which claims to fight, which to fix, and what the legal precedent looks like before the window closes. If you are managing a roster and want to see what that looks like, sign up for early access at clearverse.ai. If this is useful to someone running a creator agency or a monetized channel, pass it along.

See how ClearVerse works!

Most of them are legitimate. A creator uses a song without a license, Content ID matches it, the rightsholder monetizes the video. That’s the system working as intended.

But a growing percentage of those claims are not legitimate. They’re filed by companies and individuals who don’t own the content they’re claiming. They file anyway because the economics work in their favor: if nobody disputes, they get paid. If someone does dispute, they lose nothing.

WatchMojo, one of the largest YouTube channels with over 20 million subscribers, spent a decade tracking this problem. Their estimate: rightsholders unlawfully claimed over $2 billion between 2014 and 2019.

And that was before Content ID volume exploded. YouTube processed 2.2 billion claims in 2024 alone. Even if only a fraction of a percent are fraudulent, that’s millions of claims siphoning revenue from creators who did nothing wrong.

How Fraudulent Claims Work

Content ID is an automated system. YouTube scans uploaded videos against a database of reference files provided by verified rightsholders. If the system detects a match, it files a claim automatically. No human reviews it before the claim goes live.

The rightsholder then has options: block the video, track viewership, or monetize it. Over 90% choose to monetize. The video stays up. Ads run. Revenue flows to whoever filed the claim.

Creators can dispute. But most don’t. According to YouTube’s own transparency data, 99.5% of Content ID claims go undisputed. That’s partly because creators assume the claim must be valid. It’s also because disputing takes time, requires understanding the process, and carries risk if you’re wrong.

For anyone willing to file bad-faith claims, this creates an asymmetric opportunity. File thousands of claims. Most go undisputed. The ones that do get disputed, you release without consequence. Meanwhile, revenue accumulates from every undisputed claim.

The claimant gets 30 days to respond to a dispute. If they do nothing, the claim eventually expires. If the creator doesn’t dispute within 5 days of the claim landing, any revenue generated before the dispute is filed goes to the claimant permanently. That first week of monetization? Gone.

And here’s what makes it worse: 65% of disputed claims resolve in the creator’s favor. That means the majority of creators who fight back were right all along. But by the time they win, the video’s peak earning window has passed.

The MediaMuv Case

This isn’t theoretical. The largest documented Content ID fraud resulted in federal prison sentences.

Between 2017 and 2021, two men ran a company called MediaMuv. They convinced a YouTube partner that they owned the rights to over 50,000 songs. They didn’t own any of them. The songs were primarily Latin American music that hadn’t yet been monetized on YouTube. MediaMuv uploaded reference files, claimed the content, and collected the royalties.

For four years, the scheme generated over $23 million. The artists who actually created the music received nothing.

By 2019, MediaMuv was branching out. They started claiming soundtracks from video games, including music from Doki Doki Literature Club. YouTubers who made gameplay videos or used the game’s music in their content suddenly found their videos claimed by a company they’d never heard of.

A Twitter account called @fuckmediamuv started tracking the scheme in 2018, warning other creators. The account posted photos of the defendants with graphics of cartoon burglars, calling them “ladrones de contenido en YouTube.” YouTube content thieves.

It took until November 2021 for federal authorities to arrest them. A grand jury indicted both men on 30 felony counts including wire fraud, money laundering, and aggravated identity theft. In 2023, Jose Teran was sentenced to 70 months in prison. Webster Batista Fernandez received 46 months. Teran forfeited a house in Phoenix, a Tesla, a BMW i8, and bank accounts containing over a million dollars.

The Department of Justice called it “one of the largest music-royalty frauds ever perpetrated.”

But here’s the part that should concern every creator and agency: MediaMuv operated for four years before anyone stopped them. Four years of claims. Four years of revenue flowing to the wrong people. And that was just one operation that got caught.

Why Agencies Are Exposed

If you’re managing YouTube creators, this is your problem at two points in the workflow.

During post-production: The editor finishes a cut. Before export, you have no system to analyze the audio and flag potential issues. YouTube will scan it after upload, but by then the video is supposed to go live. If it gets flagged, someone has to go back to the editor (often in a different time zone), re-edit, re-export, re-upload to Google Drive, re-download, re-upload to YouTube, re-add thumbnails and titles. That loop burns hours and delays launch.

After a claim lands: A creator gets hit with a claim from a company you’ve never heard of. They come to you asking what happened. Is it real? Should they dispute? What’s the process? What happens if they’re wrong?

Most of the time, you don’t have good answers.

You can look up the claimant, but that doesn’t tell you whether they actually own the content. You can tell the creator to dispute, but neither of you knows whether the claim will hold up. You can wait 30 days to see if the claimant releases, but by then the video’s first week of revenue is already gone.

And if you’re managing 30 or 50 creators, this happens more than once. A claim lands on Monday. Another on Thursday. A third the following week. Each one triggers the same workflow: figure out if it’s real, decide whether to fight, file the dispute, wait, follow up.

That’s time your ops team spends playing defense instead of driving growth.

The larger issue is that you have no system for knowing which claims are worth fighting. Content ID tells you a claim was filed. It tells you who filed it. It doesn’t tell you whether that company has a history of bad-faith claims. It doesn’t tell you whether other creators have successfully disputed similar claims. It doesn’t give you any legal context for whether the claim is likely to hold up.

The Real Issue

YouTube tells you a match exists. It doesn’t tell you what it means.

When a claim lands, YouTube’s interface shows you the claimant, the content matched, and the timestamp. It gives you four options: do nothing, erase the audio, replace it with YouTube’s library, or dispute.

What it doesn’t give you: any intelligence about whether this claim is legitimate. Any history of this claimant’s dispute rate. Any legal precedent suggesting whether the claimed content is actually protected or whether you have a fair use defense. Any data point that would help you make an informed decision.

The same gap exists during editing. YouTube will scan your video after you upload it. But what if you could scan it during post-production, before the file ever touches YouTube, and get deeper analysis than YouTube provides? Not just “this matches Track X” but “this segment has high similarity to a track with an aggressive claimant history, courts have ruled against similar usage in 3 out of 4 comparable cases, here’s your risk level and here’s the recommended action.” Your editor could fix it in the timeline before export, not after upload.

Creators are expected to navigate this alone. Agencies are expected to figure it out across dozens of channels with no tools designed for the job.

This is why fraudulent claimants get away with it. The system assumes every claim is valid until proven otherwise. The burden of proof falls on the creator. And most creators don’t have the time, knowledge, or resources to fight back.

The 65% win rate on disputes tells you something important: the majority of creators who fight are right. But the 99.5% of claims that go undisputed tells you something else: most creators never try.

That gap is where money disappears.

What We’re Building

This is why ClearVerse exists.

We’re building copyright intelligence that YouTube doesn’t provide, whether your team catches an issue during editing or after upload.

During post-production: Analyze content before it’s exported and uploaded. Give editors and content teams deeper analysis than YouTube will ever provide. Not just “this matches Track X” but defensibility assessment, claimant reputation, and recommended action. Fix it in the timeline, not after the video was supposed to go live.

After a claim lands: Decision intelligence. Should you fight or fix? What does this claimant’s history look like? What have courts ruled in similar cases? AI-generated dispute responses backed by legal precedent.

For agencies managing creator rosters: Portfolio-level visibility into claims across your entire operation. Identify which claimants file bad-faith claims so you can respond with evidence, not hope. Standardized workflows so your ops team isn’t reinventing the process every time a claim lands.

YouTube detects matches. ClearVerse analyzes risk. There’s a difference.

Managing YouTube creators and seeing claims you don’t know how to evaluate? I’d like to show you what copyright intelligence looks like, before and after upload.

See how ClearVerse works!

Have you dealt with a claim that felt fraudulent? How did you handle it? I’m curious what’s working and where you’re seeing gaps.

— Christian

P.S. The MediaMuv defendants are in prison, but the incentive structure that made their scheme possible hasn’t changed. As long as undisputed claims pay out and bad-faith claimants face no consequences until someone sues, this problem will keep growing. Forward this to anyone managing creator portfolios who should be paying attention.

This week, users demonstrated how easy those filters are to bypass. Upload a YouTube cover version of a copyrighted song instead of the original recording and Suno’s detection doesn’t flag it. Change a few words to homophones and the lyric filter lets it through. One tester bypassed the system on Beyoncé’s “Freedom” by changing “rain on this bitter love” to “reign on.” Beyond the first verse and chorus, they didn’t need to make any changes at all.

For a company facing up to $150,000 in statutory damages per infringed work in federal court, that’s a problem.

But here’s what caught my attention: when those filters fail, Suno faces a lawsuit. Your creator faces a Content ID claim. And you face the lost revenue, the ops headache, and the conversation about why you didn’t catch it before upload.

The filters protect Suno. They don’t protect your creators.

What’s Actually Happening

The bypass methods aren’t sophisticated. They’re embarrassingly simple.

Suno scans for copyrighted material in two places: uploaded audio and typed lyrics. The audio filter checks uploads against known recordings. The lyric filter matches text against protected song words.

Both have gaps.

The audio filter only recognizes original recordings. A YouTube cover of a song doesn’t trigger it. The AI then uses that cover as a style reference and produces output that sounds like the copyrighted original without technically matching it. Users have been sharing this workaround openly. Upload a cover, get the vibe, bypass the filter.

The lyric filter relies on exact text matching. Homophones work. Phonetic spelling works. Small changes defeat it. Swap “sweet” for “suite” and you’re through. The filter scans for the copyrighted text string; if it doesn’t match exactly, it passes.

Independent artists testing Suno’s v5 model found their own songs cleared the copyright filter with zero modifications. Tracks by smaller artists, self-distributed through Bandcamp or DistroKid, slipped through without any changes at all. The protection appears strongest for major-label catalogs and weakest for everyone else.

That asymmetry matters. If you’re managing creators who use AI music, the tracks most likely to trigger Content ID claims might be the ones Suno’s filters are least likely to catch.

The Legal Context

Suno isn’t operating in a vacuum. It’s in the middle of active litigation.

The RIAA filed suit on June 24, 2024, in the US District Court for the District of Massachusetts on behalf of Universal, Sony, and Warner. The complaint alleges mass infringement, accusing Suno of training its AI on copyrighted recordings without permission. It seeks up to $150,000 per infringed work, plus $2,500 for each act of circumventing encryption protections.

In its August 2024 answer, Suno argued fair use. In the same filing, Suno acknowledged that its training data “presumably included recordings whose rights are owned by the Plaintiffs.” The company declined to disclose what was actually in that data, calling it “confidential business information.”

That fair use argument took a hit in May 2025. The US Copyright Office released its Part 3 report on generative AI training and concluded that using copyrighted expressive works to generate competing content goes “beyond established fair use boundaries.” The report emphasized this is especially true when the AI produces outputs that substitute for the originals in the market.

Warner Music Group subsequently settled and formed a licensing partnership with Suno. The deal announced in November 2025 includes new “licensed models” rolling out in 2026, download restrictions for free users, and monthly download caps for paid subscribers. Financial terms weren’t disclosed.

But Universal and Sony remain in active litigation. And a separate class action by independent artists is also pending in the same Massachusetts court, with an amended complaint filed September 22, 2025. That filing tripled the original complaint’s length and added stream-ripping allegations, claiming Suno bypassed YouTube’s encryption to harvest training data.

Internationally, GEMA — Germany’s version of ASCAP or BMI, representing over 95,000 composers and songwriters — became the first music rights organization worldwide to sue an AI music generator when it filed against Suno in January 2025. Their tests allegedly produced AI outputs for songs including “Daddy Cool,” “Mambo No. 5,” “Forever Young,” and “Atemlos” that matched the originals in melody, harmony, and rhythm without specifying those elements in the prompt. A ruling is scheduled for June 12, 2026.

The legal landscape is moving. But none of that helps your creator when a claim lands tomorrow.

Why This Is an Agency Problem

Suno’s legal exposure is Suno’s problem. Your operational exposure is yours.

When a creator uses AI-generated music in a YouTube video and that music triggers a Content ID match, YouTube doesn’t care whether Suno’s filters should have caught it. Content ID scans the final audio. If it matches a reference file in YouTube’s database, a claim gets filed. Revenue redirects to the claimant. The yellow icon appears. And your creator wants to know what happened.

You can’t see it coming. AI music platforms don’t tell creators “this track might trigger a claim.” They tell creators the filters are working. Your ops team has no visibility into whether a creator’s AI-generated track shares enough characteristics with protected content to trigger a match.

Content ID doesn’t care about source. It uses acoustic fingerprinting. It doesn’t distinguish between music created in a DAW, pulled from a library, or generated by an AI that trained on copyrighted material. If the output sounds similar enough to a reference file, you get a claim. The fact that Suno said it was safe doesn’t matter.

And if you’re managing 30 creators across your roster who are using various AI tools with varying filter quality, that’s cumulative risk you can’t currently track. One creator experimenting with Suno is a curiosity. A dozen creators generating tracks with tools whose filters have documented gaps is a portfolio-level exposure you have no system to manage.

There’s also the ownership question, which most agencies aren’t thinking about yet. Under current Copyright Office guidance, music generated entirely by AI without meaningful human authorship may not be copyrightable. That means creators using raw AI output might not even own what they’re uploading. If a dispute arises over rights to the track, the answer might be: nobody owns it.

The time cost adds up too. Every claim triggers a workflow. Someone on your team notices the claim, assesses whether it’s legitimate, decides whether to dispute, files the dispute, monitors it for 30 days, follows up if it gets rejected. Multiply that by six or eight claims per month across your roster and you’ve burned hours your ops team could have spent on growth, creator support, or closing brand deals.

And every claim erodes trust. Creators see the yellow icon. They watch their AdSense estimate drop. They ask what happened, why you didn’t catch it, what you’re doing to prevent it next time. You don’t have a good answer because there’s no tool that catches it before upload. The agencies that retain their best talent will be the ones who can say: “We checked it before you uploaded. It’s clean.”

The Real Issue

The filter bypass story exposes something that was already true.

AI music platforms position their filters as protection for creators. That’s not quite right. The filters protect the platform from the most obvious liability. They reduce the chance that Suno generates something that sounds exactly like “Bohemian Rhapsody” and triggers immediate legal action. They don’t protect creators from Content ID claims. They don’t guarantee the output is safe to monetize. They don’t mean you won’t get a claim.

Suno’s Terms of Service make this explicit if you read them. The platform grants users certain rights but disclaims any guarantee that content is free of third-party claims. The legal risk of using the output sits with the user, not the platform.

For agencies, relying on platform filters isn’t a risk management strategy. It’s hope dressed up as process.

What We’re Building

This is why ClearVerse exists.

We analyze audio before it goes live. Not after a claim lands. Before your creator hits publish. Our system scans against known Content ID reference patterns and scores risk based on legal precedent, not just acoustic similarity. If a track is likely to trigger a claim, you know before you lose the first 48 hours of revenue.

The platforms protect themselves. We help you protect your creators.

Managing YouTube creators who use AI-generated music? I’d like to show you what pre-publish protection looks like.

See how ClearVerse works!

Are your creators using AI music tools? What’s your current process for checking whether those tracks are safe? I’m curious what’s working and what gaps you’re seeing.

— Christian

P.S. Suno’s filters might improve. The lawsuits will eventually resolve. But the fundamental problem remains: platform filters protect platforms, not creators. Forward this to anyone managing creator portfolios who should be thinking about this.

You check your dashboard. One of your creators published yesterday. The video is performing. Views are climbing. And then you see it: a copyright claim landed six hours after upload. The monetization icon is yellow. The revenue from the video’s best 48 hours is already redirected to whoever filed the claim.

You didn’t see it coming. Neither did your creator. And now you’re in reactive mode.

This is happening across the industry at massive scale. YouTube processed 2.2 billion Content ID claims in 2024. That’s 6 million claims per day, over 4,000 per minute. In more than 90% of cases, rightsholders chose to monetize the claimed content rather than block it.

The videos stay up. The views keep coming. The ads run. But the revenue flows to the claimant, not to your creator, and not to you.

Why the First 48 Hours Matter to Your Business

YouTube’s algorithm tests new content aggressively in the first 24 to 48 hours. Strong early performance triggers broader distribution. This is also when ad revenue concentrates. A video that reaches 100,000 views in its first week will often generate the majority of those views in the first two days.

For agencies, this creates a timing problem. A claim that lands on day one captures the revenue from the video’s highest-earning window. Even if the claim is eventually released or disputed successfully, that window is already closed.

YouTube gives claimants 30 days to respond to disputes. If your team doesn’t file within 5 days of the claim, any revenue generated before the dispute is filed goes to the claimant permanently. By the time any dispute resolves, the algorithm has already moved on.

Here’s the part that should concern you: over 65% of disputed claims eventually resolve in the uploader’s favor. The claimant either releases the claim or doesn’t respond. That means the majority of disputes involve claims that shouldn’t have stuck, but the revenue from the video’s peak performance window is still gone.

The Real Cost to Your Agency

The direct revenue hit is straightforward to calculate. Take a creator on your roster in a finance or tech vertical. One video hits 100,000 views in its first week. At a $10 CPM, that’s $1,000 gross. YouTube takes 45%, leaving $550 for the creator. Your 20% management fee on that is $110. A single claim wipes out both numbers.

Multiply that across your roster. If you’re managing 30 creators publishing weekly, that’s 120+ uploads per month flowing through your portfolio. Even a small claim rate means thousands in lost revenue per month, revenue your creators expected and revenue your agency counted on.

But the revenue loss is only the most visible cost.

Ops burden. Every claim triggers work for your team. Someone has to monitor for claims across your roster, assess each one, decide whether to dispute, file the dispute, track it for 30 days, and follow up if it’s rejected. At scale, this becomes a significant time sink. Hours your team spends on claim management are hours they’re not spending on growth, brand partnerships, or creator support.

Creator trust. When a creator sees their AdSense estimate drop to zero, they notice immediately. The next message you get is some version of: “What happened? Why didn’t you catch this? What are you doing to prevent it next time?” Every claim is a conversation that erodes confidence. Creators with options will eventually move to agencies that can protect their revenue.

Portfolio drag. Creators who get burned by claims often become conservative. They avoid music that would make their content more engaging. They stop experimenting with formats that might trigger claims. Over time, this shows up as slower growth across your roster, less competitive content, and creators who underperform their potential.

The Visibility Gap

The core problem isn’t that claims happen. It’s that you have no way to see them coming and no system to manage the risk across your portfolio.

Content ID is reactive. It scans content after upload. By the time a claim appears in YouTube Studio, the video is already live, the algorithm is already testing it, and the first 48 hours have already started.

There’s no portfolio-level view of copyright risk before content goes out. No system that flags which uploads are likely to trigger claims. No way to know whether the background track a creator is using will be safe or not.

Most agencies handle this in one of two ways. Some mandate royalty-free music libraries and build that into creator contracts. That solves the claim problem but limits what your creators can produce. Others take a reactive approach: deal with claims when they happen, dispute when possible, and accept the losses.

Neither scales well. The first constrains your creators. The second burns money and ops time you never recover.

The agencies that retain top talent and protect their margins will be the ones that can tell creators, before they hit publish: “We checked it. It’s clean.”

What Leading Agencies Are Building

The agencies getting ahead of this are building process, but most of what they’re doing is manual.

Some require creators to submit audio for review before upload. That works when you’re managing 10 creators. It breaks at 50+. The review backlog becomes its own bottleneck.

Others build internal documentation: wikis of “safe” tracks, lists of music to avoid, playbooks for handling claims. This works until the person who built it leaves. Then the institutional knowledge walks out the door.

The most sophisticated agencies layer licensed music platforms, internal training, and dedicated ops capacity for claim management. But even this is patchwork. None of it gives you real-time, portfolio-wide visibility into risk before content goes live.

That’s the gap ClearVerse is built to fill.

What Portfolio-Level Copyright Protection Looks Like

ClearVerse gives your agency a single dashboard for copyright risk across your entire roster.

Before content goes live, your team can scan audio against known Content ID reference patterns. We score risk based on legal precedent, not just acoustic similarity. You see which uploads are likely to trigger claims before they publish.

When something flags, you have options. Swap the track. Clear the license. Or make an informed decision that the risk is acceptable. Either way, you’re not finding out after the damage is done.

For your ops team, this means fewer surprise claims, fewer disputes to manage, and fewer difficult conversations with creators. For your business, it means protecting the revenue your roster generates and the margin your agency depends on.

If you’re running a creator agency and want to see what portfolio-level copyright protection looks like, I’d like to show you.

See how ClearVerse works!

How does your agency handle copyright risk across your roster today? I’m curious what’s working, what’s not, and what you wish you had. Reply or drop a comment below.

— Christian

P.S. If this was useful, forward it to another agency founder or ops lead. The more agency operators understand what reactive claim management actually costs, the faster this space moves toward something better.

If you’ve been following the AI music space, another lawsuit might not feel like news. Suno and Udio have been fighting the major labels for over a year. The same indie artists suing Google already have pending cases against both platforms.

But this case is structurally different. The legal claims are similar: copyright infringement, unlicensed training data, the usual. What’s different is who Google is and what Google controls.

The plaintiffs’ core argument, laid out in the opening pages of their 118-page complaint:

“Google owns the platform where independent musicians distribute their music. Google runs the system that identifies who owns it. And then Google used both to train a product that competes with the very artists who trusted it with their work.”

That’s not a startup scraping the web. According to the lawsuit, it’s the owner of YouTube using the same infrastructure artists depend on to build something that competes with them.

Whether that argument holds up legally is an open question. But it’s worth understanding what’s being alleged and why it matters for creators.

What Happened

Google launched Lyria 3 on February 18 through its Gemini app. The tool generates 30-second tracks with vocals and lyrics from text prompts or images. It’s now available to more than 750 million monthly active Gemini users worldwide.

In its launch announcement, Google said it developed the technology “responsibly in collaboration with the music community” and has “been very mindful of copyright and partner agreements.”

The plaintiffs dispute that framing.

Their complaint points to Google’s own published research, specifically peer-reviewed papers authored by Google researchers, that describe the training data in detail.

A 2022 paper describes collecting approximately 50 million internet music videos, extracting 30-second clips from each, and retaining roughly 44 million clips totaling nearly 370,000 hours of recorded music.

A 2023 paper describes training on 5 million clips totaling 280,000 hours of audio.

According to the lawsuit: “Both papers were published in peer-reviewed venues under the names of Google researchers. Neither mentions a license. Neither discusses a consent mechanism. Neither identifies a single rights holder whose permission was sought.”

The plaintiffs argue Google “has not named any agreement. It has not specified a single license covering training data.”

Google hasn’t publicly responded to the lawsuit yet.

Why the Plaintiffs Say This Case Is Different

When Suno and Udio got sued, the lawsuits targeted startups that allegedly scraped music from the internet. Both companies have argued fair use. The labels argued mass infringement. Courts are still deciding.

The plaintiffs argue Google’s situation is different because of what Google already controls.

Google owns YouTube, one of the largest repositories of recorded music in the world. Artists upload their work to YouTube for distribution, discovery, and monetization.

Google also runs Content ID, the automated rights management system that scans uploads and matches them against reference files. The system reportedly contains more than 50 million reference files submitted by rights holders.

The lawsuit argues this gave Google unique access. It didn’t need to scrape the web because it already had the music. It didn’t need to guess who owned what because Content ID already tracked ownership.

As the complaint puts it, Google had “every opportunity to develop this product legally. It has the technical infrastructure, financial resources, and industry connections to clear rights before training.”

The plaintiffs allege Google “chose not to do so, not because licensing was impossible, but because copying was faster and cheaper.”

That’s their argument. Whether courts agree is a different question.

The DeepMind Connection

One detail worth noting: Udio, the AI music platform that settled with Universal and Warner last year, was founded by former Google DeepMind researchers.

The four co-founders who built Udio’s core technology previously worked on generative music at Google. According to the lawsuit, “Within months, all four left Google DeepMind, formed a new company, and publicly launched Udio, another AI music-generation service trained on copyrighted recordings without authorization.”

Google launched Lyria 3 after Udio’s settlements were announced.

Days after that launch, Google acquired ProducerAI, the AI music platform formerly known as Riffusion, bringing the team into Google Labs.

The plaintiffs see a pattern. Whether it’s legally relevant is something courts will determine.

What the Lawsuit Alleges

The complaint lists 16 alleged violations under federal and state law:

Copyright infringement on both sound recordings and musical compositions. This is the core claim: that Google copied millions of works without permission.

Removal of copyright management information. The lawsuit alleges Google stripped identifying metadata during the training process.

False endorsement under the Lanham Act. When users create with Lyria 3, Google identifies them as the “creator.” The plaintiffs argue this generates false attribution for works allegedly derived from unlicensed material.

Illinois BIPA violations. Google’s privacy notice for ProducerAI mentions that it “may involve the extraction of a biometric voiceprint.” The plaintiffs argue this could violate Illinois biometric privacy law if done without consent.

The case is structured as a proposed class action. The named plaintiffs, indie artists from New York, Los Angeles, Atlanta, and Chicago, are seeking to represent all independent artists whose work may have been used.

What Google Has Said

Google hasn’t formally responded to the lawsuit.

At the Lyria 3 launch, the company said it trains on music it has “a right to use under our terms of service, partner agreements, and applicable law.”

The plaintiffs argue that language is vague. Being “mindful” of copyright isn’t the same as having licenses. “Partner agreements” could mean formal licensing deals or standard terms-of-service language.

If Google has specific licensing agreements covering the training data, it hasn’t publicly disclosed them.

What This Means for Creators

If you’re using AI music tools or releasing music on streaming platforms, a few things are worth noting:

The scale is significant. Gemini has 750 million monthly active users. Compare that to Suno’s roughly 100 million total users or Udio’s smaller base. If Lyria 3 becomes a go-to tool for casual music creation, the volume of AI-generated content entering the ecosystem will grow substantially.

The platform relationship question is sharpening. This lawsuit forces a question that’s been in the background: what do platforms owe the creators who use them? Artists upload to YouTube expecting distribution, not training data extraction. Whether that expectation is legally enforceable is what courts will decide.

Legal uncertainty remains. Google is being sued under similar theories to the Suno and Udio cases. But those cases haven’t been decided on fair use. The Copyright Office is still developing guidance. If you’re releasing AI-assisted music, the rules aren’t fully clear yet.

What to Watch

Whether Google settles or fights. The major labels have been settling with AI platforms rather than going to trial. Google has far more resources than any AI music startup. It could fight through trial and potentially establish precedent, or it could settle quietly.

Class certification. This is a proposed class action. If the class gets certified, the potential damages grow significantly. If it doesn’t, this remains a smaller case.

Whether the major labels file their own suit. Right now, this is an indie artist lawsuit. The RIAA hasn’t filed against Google. If they do, the dynamics change; Google has existing relationships with all three majors through YouTube licensing and Content ID.

The Larger Question

This lawsuit sits at an interesting intersection.

YouTube has been controversial in the music industry for years. Artists have debated the royalty rates, the value gap, and platform power. But YouTube remained essential for reach and discovery.

This case asks whether the calculation changes when the platform you depend on is also training AI that could compete with you.

That’s not a legal question with a clear answer. It’s a relationship question that creators and platforms will navigate over the coming years, regardless of how this lawsuit resolves.

What We’re Building

At ClearVerse, we’re building tools to help creators understand copyright exposure before release, not after a claim arrives.

Cases like this add complexity to an already complicated landscape. Understanding where your music goes, who might be using it, and what rights you actually have is becoming more important, not less.

If that resonates, you can sign up for early access at clearverse.ai.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

What’s your read on this lawsuit? Does it change how you think about uploading to platforms? I’d like to hear your perspective in the comments.

P.S. If this helped clarify what’s happening with AI music and copyright, send it to a creator friend. These issues affect everyone making music right now.

This wasn’t a dramatic ruling. The Court simply refused to take up the case, leaving lower court decisions intact. But that procedural silence carries enormous weight for anyone creating music with AI tools.

If you use Suno, Udio, or any other AI music generator, this decision directly affects the legal foundation of your work. Let’s break down what actually happened, what it means, and what you need to do about it.

The Case That Finally Reached (and Was Rejected by) the Highest Court

Dr. Stephen Thaler is a computer scientist from St. Charles, Missouri, and founder of Imagination Engines Incorporated. In 2012, his AI system DABUS (also called the “Creativity Machine”) generated a visual artwork titled “A Recent Entrance to Paradise.”

In November 2018, Thaler filed for copyright registration; but with an unusual twist. He listed DABUS, the AI, as the author, describing the work as “created autonomously by machine.” He sought to own the copyright as a “work-for-hire to the owner of the Creativity Machine.”

The Copyright Office rejected his application in August 2019, stating: “We cannot register this work because it lacks the human authorship necessary to support a copyright claim.”

Thaler didn’t give up. He filed reconsideration requests in September 2019, May 2020, and February 2022. Each was rejected for the same reason.

In June 2022, he sued the Copyright Office. In August 2023, Judge Beryl A. Howell of the U.S. District Court for D.C. ruled against him, writing what has become the defining phrase of this legal era:

“Human authorship is a bedrock requirement of copyright.”

Thaler appealed. On March 18, 2025, the D.C. Circuit Court of Appeals unanimously affirmed the district court’s decision. Judge Patricia Millett’s opinion was unambiguous: the Copyright Act of 1976 “requires all eligible work to be authored in the first instance by a human being.”

The court even offered an analogy: AI should be understood as a sophisticated tool, “akin to a laboratory instrument.” The tool might help create something, but it cannot be the author.

Thaler petitioned the Supreme Court in October 2025, arguing the case was of “paramount importance” given the rapid rise of generative AI. On January 23, 2026, the Trump administration’s Department of Justice filed a brief urging the Court to decline the case, noting: “Although the Copyright Act does not define the term ‘author,’ multiple provisions of the act make clear that the term refers to a human rather than a machine.”

On March 2, 2026, the Supreme Court denied certiorari without comment.

Thaler’s attorneys responded: The decision risks “irreversibly and negatively impacting AI development and use in the creative industry during critically important years.”

Maybe. But the law is now settled; at least until Congress acts.

What This Actually Means (And What It Doesn’t)

Let’s be precise about what this decision does and doesn’t do.

What it confirms:

Works created entirely by AI, without meaningful human creative contribution, cannot be copyrighted under U.S. law. If you type a prompt into Suno and release the output unchanged, that track likely has no copyright protection. Anyone could copy it, and you’d have limited legal recourse.

What it does NOT do:

This ruling doesn’t ban AI-assisted music creation. The D.C. Circuit explicitly stated that “the human authorship requirement does not prohibit copyrighting work made by or with the assistance of artificial intelligence.”

It doesn’t affect your Suno or Udio subscription. Those platforms grant you commercial use rights through their terms of service; a contractual license, not a copyright transfer. You can still monetize AI-generated tracks on YouTube, Spotify, or elsewhere.

It doesn’t address the ongoing lawsuits over AI training data. Whether companies like Suno can legally train on copyrighted music is a separate legal question (fair use) from whether their outputs are copyrightable.

The Copyright Office Already Told Us This Was Coming

If you’ve been following this space, the Supreme Court’s decision shouldn’t surprise you. The Copyright Office has been consistent for years.

On January 29, 2025, the Office released Part 2 of its Copyright and Artificial Intelligence Report, which stated clearly:

“Given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output.”

The report drew a crucial distinction: AI can be used “to assist” human creativity, but not as a “stand in” for it. Using AI to help brainstorm, clean up audio, or generate reference ideas? That’s fine. Typing a prompt and releasing the raw output as your creative work? That likely won’t qualify for copyright protection.

Register of Copyrights Shira Perlmutter put it plainly: “Where human creativity is expressed through the use of AI systems, it continues to enjoy protection. Extending protection to material whose expressive elements are determined by a machine, however, would undermine rather than further the constitutional goals of copyright.”

What This Means for You as a Creator

Here’s the practical reality for anyone using AI music tools in 2026:

1. Raw AI output = legally vulnerable

If you generate a track in Suno and release it unchanged, you have commercial use rights through your subscription; but likely no copyright. If someone copies your track, your legal standing to enforce is weak.

This is why Suno’s own terms of service include this disclaimer: “Due to the nature of machine learning, Suno makes no representation or warranty to you that any copyright will vest in any Output.”

They’re telling you this directly.

2. Human contribution strengthens your position

The more you add, the stronger your claim. According to the Copyright Office’s guidance, copyright can attach to:

• Human-authored lyrics (even if set to AI-generated music)

• Substantial modifications to AI output

• Creative arrangement or selection of AI-generated elements

• Original vocals, instruments, or production you layer on top

If you’re treating AI as raw material and putting your creative stamp on it, you’re in a much better position than someone just hitting “generate” and “export.”

3. Documentation matters

As AI becomes more embedded in creative workflows, you may need to demonstrate what you contributed. Save your drafts. Keep notes on your process. Document which elements are yours versus AI-generated.

This isn’t paranoia; it’s prudent practice in a landscape where the legal status of your work may be challenged.

4. Platform disclosure requirements are here

Spotify, Apple Music, and YouTube are implementing or expanding disclosure requirements for AI-generated content. Spotify’s DDEX standard requires labels and distributors to indicate where AI was used; in vocals, instruments, or production. YouTube treats “raw” AI audio with minimal human input as low-value content, potentially ineligible for monetization.

Bandcamp went further in January 2026, banning AI-generated music entirely. Their stated goal: ensuring “fans have confidence that the music they find on Bandcamp was created by humans.”

If you’re distributing AI-assisted music, disclosure isn’t optional anymore. And non-disclosure could cost you.

The Bigger Picture

The Supreme Court’s refusal to hear Thaler’s case doesn’t mean AI is incompatible with copyright. It means AI cannot replace human authorship.

This is an important distinction. Courts and the Copyright Office aren’t hostile to creators who use AI tools; they’re drawing a line around works that have no meaningful human creative input.

For YouTube creators using AI music in their videos, the practical impact is limited. Your videos are your creative work. Background music, whether licensed, AI-generated, or from the YouTube Audio Library, is just one element.

But if you’re building a music-focused brand, releasing tracks on streaming platforms, or otherwise treating AI-generated music as a primary creative output, you need to think carefully about:

• What human contribution are you making?

• Can you document that contribution?

• Are you disclosing AI use where required?

• Do you understand the difference between commercial use rights and copyright ownership?

Where This Goes From Here

The D.C. Circuit acknowledged that AI technology will continue to evolve. Judge Millett’s opinion even referenced Star Trek’s Data as an example of a future AI that might respond to economic incentives; noting that “if the human authorship requirement stymied the creation of original work,” Congress could address it.

That door isn’t closed forever. But for now, Congress hasn’t acted, and courts have spoken: if you want copyright protection, there must be a human in the creative process.

For creators, this creates both risk and opportunity. The risk is obvious: raw AI output is legally thin. The opportunity is that human creativity still matters; and the law says so.

Before You Hit Publish

If you’re using AI music tools, ask yourself:

1. What am I actually contributing to this track?

2. Is my contribution substantial enough to demonstrate authorship?

3. Have I disclosed AI use where my distribution platform requires it?

4. Do I understand what my platform license does and doesn’t give me?

These aren’t just legal questions; they’re questions about the kind of creator you want to be.

AI can be an incredible tool for music creation. But tools don’t create alone. The law just confirmed what many of us already believed: the human behind the machine still matters.

ClearVerse is building technology to help creators understand the legal exposure in their music before they release it — so you can create with confidence, not confusion.

Signups are open. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

What’s your take on the Supreme Court’s decision? Does this change how you think about AI music tools? I’d love to hear your perspective in the comments.

P.S. If this helped clarify the copyright landscape, forward it to a creator friend who uses AI tools. The more we understand the rules, the better we can work within them.

Except it didn’t; because the episode revealed something important that every creator using AI music tools needs to understand.

What Actually Happened

Let me walk you through the timeline, because the details matter.

November 2025: Suno’s Help Center was clear. If you paid for Pro or Premier, “you own the songs” and “you are granted a commercial use license to monetize those songs.” Simple.

November 25, 2025: Suno announced a landmark settlement with Warner Music Group, ending the label’s copyright lawsuit and launching a licensing partnership. The deal promised “more advanced and licensed models” coming in 2026.

Late December 2025: Creators started noticing that Suno’s FAQ language had quietly changed. The new version read differently: “Even with granted commercial use rights, you generally are not considered the owner of the songs, since the output was generated by Suno.”

Another page went further: “The answer (as to whether Suno owns outputs) is a bit complicated, but generally, yes. Suno is ultimately responsible for the output itself, though you help guide it.”

That’s a meaningful shift from “you own the songs” to “you’re not considered the owner.”

January 7, 2026: After creator feedback spread across Reddit, Discord, and social media, Suno posted a clarification on X: “We recently made an update to our Knowledge Base (FAQ) and it didn’t clearly describe ownership on our platform... you do own the Output you make in the Pro and Premier plans and you are granted commercial rights to use those Outputs.”

Suno reverted the FAQ to its previous version. The company emphasized that the Terms of Service, the actual legal document, hadn’t changed since November 6, 2025.

Why This Matters More Than It Seems

Here’s what most coverage of this story missed: even with the clarification, there’s a critical distinction buried in Suno’s Terms of Service that every creator should understand.

The Terms say Suno “assigns to you all of its right, title and interest” in outputs you create while subscribed to a paid plan. That sounds like ownership. And according to Suno, it is.

But the very next sentence adds: “However, due to the nature of machine learning, Suno makes no representation or warranty to you that any copyright will vest in any Output.”

Read that again. You own the output. But that output might not be copyrightable.

This isn’t Suno being sneaky. It’s Suno being honest about a genuinely unsettled area of law.

The Difference Between Ownership and Copyright

This is where most creators get confused; and it’s the most important thing to understand if you’re using AI tools to create music.

Ownership is about your relationship with the platform. Suno says paid subscribers own their outputs. That means Suno won’t claim your songs, won’t take a cut of your revenue, and won’t stop you from distributing them.

Copyright is about your relationship with the law. Under current U.S. Copyright Office guidance, works created entirely by AI, without sufficient human creative input, generally aren’t eligible for copyright protection.

You can own something that isn’t copyrightable. You just can’t stop someone else from copying it.

Here’s what that looks like in practice:

• You create a song with Suno while subscribed to Pro

• You own that song (per Suno’s terms)

• You upload it to Spotify through DistroKid

• You earn royalties from streams

• Someone else downloads your song and re-uploads it as their own

• You have limited legal recourse because the song may not be copyrightable

This isn’t a Suno problem. It’s an AI music problem. Udio’s terms have similar disclaimers. So will every AI music platform until the law catches up.

What the December Confusion Actually Revealed

The brief language change, and the speed of the backlash, tells us something important about where AI music is heading.

Suno now has 2 million paid subscribers and $300 million in annual recurring revenue. Users generate roughly 7 million tracks per day on the platform. That’s more music every two weeks than Spotify’s entire catalog.

When you have that many creators depending on your platform, words matter. The difference between “you own the songs” and “you’re granted commercial use rights” might seem like lawyer-speak, but creators immediately understood the implications.

The episode also landed at a sensitive moment. Suno had just settled with Warner Music. New “licensed models” were coming. Download caps were announced. Creators were already watching for signs that the platform might become more restrictive.

Suno moved quickly to clarify, and I think that’s worth acknowledging. Once the confusion surfaced publicly, they posted a clear statement within days and reverted the FAQ. Compare that to platforms that let ambiguity linger for months.

But the underlying tension isn’t going away. AI music platforms are caught between two forces: creators who want simple, clear ownership, and a legal landscape that can’t provide it yet.

What This Means If You’re a YouTube Creator

If you’re using Suno (or Udio, or any AI music tool) to create background music, intros, or full tracks for your content, here’s what you should actually do:

1. Understand that “ownership” has limits

You can monetize AI-generated music on YouTube, Spotify, and other platforms. But you may not be able to enforce copyright claims against others who copy it. Plan accordingly.

2. Add human creativity where it matters

The Copyright Office has indicated that human creative input can make AI-assisted works copyrightable. If you write original lyrics, perform vocals, or substantially modify the AI output, you strengthen your legal position.

3. Read the terms; especially the disclaimers

Every AI platform has language about copyright uncertainty. This isn’t fine print designed to trick you. It’s an accurate description of the current legal landscape.

4. Keep records

Document when you created tracks, what plan you were on, and what inputs you provided. If ownership questions arise later, you’ll want receipts.

5. Stay informed

The rules are changing. The Copyright Office is still developing guidance. Courts are still deciding cases. What’s true today might shift in six months.

The Bigger Picture

Suno isn’t the villain in this story. Neither are the creators who raised concerns. Everyone is navigating genuinely uncharted territory.

AI music tools are giving millions of people the ability to create songs who never could before. That’s remarkable. It’s also creating legal and business questions that don’t have clean answers yet.

The platforms are trying to build sustainable businesses while lawsuits are still pending. Creators are trying to monetize their work while copyright rules remain unclear. Labels are trying to protect their catalogs while also exploring licensing deals.

All of this is happening simultaneously, in public, at scale.

The December confusion at Suno was a small window into how fragile the current equilibrium is. A few sentences in an FAQ created genuine anxiety among creators who depend on the platform. That anxiety was reasonable; because the underlying uncertainty is real.

What We’re Building at ClearVerse

This is exactly why I started ClearVerse.

When I talk to creators using AI music tools, the same questions come up over and over: Can I monetize this? Will I get a copyright claim? What if my song sounds similar to something else? Am I actually protected?

The answers aren’t simple. They depend on how you created the music, what inputs you used, what the output sounds like, and which platform you’re releasing on.

We’re building a tool that helps you understand your risk before you release; not after you get a claim or a takedown notice. Think of it as a pre-flight check for your music.

Because the best time to understand your legal exposure isn’t when something goes wrong. It’s before you hit publish.

Signups are open. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

Have you been following the Suno ownership situation? Have questions about your rights when using AI music tools? Drop a comment — I read all of them.

P.S. If this was helpful, forward it to a creator friend who uses AI music. These questions affect everyone in the space, and most people don’t know what they don’t know.

But here’s what those headlines missed: the inability to detect AI music isn’t the crisis. The crisis is what happens after the music gets uploaded.

The Real Story Behind the 97%

Let’s be clear about what the Deezer study actually found.

Yes, 97% of people failed a blind listening test. But dig into the details and you see something more interesting: 52% of those people felt uncomfortable when they learned they couldn’t tell the difference. 71% were surprised. 80% said AI-generated music should be clearly labeled.

In other words, people don’t want to be fooled. They want to know what they’re listening to.

And when you look at what’s actually happening on streaming platforms, you start to understand why.

60,000 Tracks Per Day

Deezer is the only streaming platform that publicly reports how much AI content it receives. In January 2025, they counted 10,000 fully AI-generated tracks uploaded daily. By September, it was 30,000. By November, 50,000. By January 2026, it hit 60,000.

That’s 39% of all music delivered to the platform. Every single day.

But here’s the number that matters most: up to 85% of the streams on those AI tracks are fraudulent.

Not “some.” Not “a concerning percentage.” Eighty-five percent.

The fraud works like this: someone uses an AI tool to generate thousands of tracks. They upload them under fake artist names. Then they use bots, manipulated playlists, and click farms to rack up streams. The streaming platform’s royalty pool pays out based on streams, so money flows from legitimate artists to scammers running automation scripts.

Deezer caught this because they built detection systems. They tag AI music, exclude it from algorithmic recommendations, and demonetize fraudulent streams. Across their entire catalog, fraud accounts for about 8% of streams. But within AI-generated content specifically? It’s 85%.

The problem isn’t that AI music sounds real. The problem is that AI music is primarily being used to steal.

What the Other Platforms Are Dealing With

Deezer is relatively small compared to Spotify and Apple Music. But they’re the only ones publishing detailed AI upload data.

What we know about the larger platforms:

Apple Music’s VP Oliver Schusser revealed in early February that they identified and demonetized 2 billion fraudulent streams in 2025. Two billion. He called it a “zero-sum game” and announced they’re doubling their fraud penalties from a maximum of 25% to 50% of would-be royalties.

Spotify removed over 75 million “spammy” tracks in 2025 alone, acknowledging that AI has made it “easier than ever for bad actors to mass upload content.” They’ve rolled out new spam filters and banned unauthorized AI voice clones.

These aren’t hypothetical concerns. In September 2024, Michael Smith was charged with wire fraud for using AI to generate hundreds of thousands of songs, then using bot accounts to stream them. At his peak, prosecutors said he was generating 661,440 fake streams per day across Spotify and Apple Music.

The streaming economy is being gamed at scale, and AI is the tool that makes it possible.

Why Detection Isn’t the Answer

Here’s where it gets complicated.

Deezer’s detection system can identify 100% AI-generated tracks from tools like Suno and Udio with 99.8% accuracy. They’ve applied for patents. They’re licensing it to other companies.

But detection only solves part of the problem. It tells you what something is. It doesn’t tell you whether it’s legal, ethical, or worth paying for.

Consider the edge cases:

A poet uses Suno to set her lyrics to music. She writes every word, but the instrumentation is AI-generated. Xania Monet, created this way, became the first AI artist to chart on Billboard and signed a multimillion-dollar record deal.

A band called The Velvet Sundown racked up 500,000 monthly Spotify listeners before anyone realized they weren’t human. The creator later admitted it was a “hoax” to see how far AI music could go.

An AI-generated country track called “Walk My Walk” hit number one on the Billboard chart in November 2025.

Some of this is creative experimentation. Some is outright fraud. The technology to detect AI doesn’t help you distinguish between the two.

What Actually Matters

The 97% headline makes for good panic. But here’s what the Deezer study actually tells us about what people care about:

80% want AI music clearly labeled.

73% of streaming users want to know if their platform is recommending AI content.

69% think payouts for AI music should be lower than for human music.

65% say it shouldn’t be allowed to use copyrighted material to train AI models.

70% believe AI music threatens musicians’ livelihoods.

People aren’t asking “can I tell the difference?” They’re asking “is this fair?”

And that’s a much harder question to answer.

The Platform Response

Right now, the platforms are responding to AI music in three ways:

Detection and labeling. Deezer tags AI content and excludes it from recommendations. This is the most aggressive approach, but it requires building and maintaining sophisticated detection systems.

Fraud enforcement. Apple Music and Spotify are demonetizing fake streams and penalizing distributors. This targets the symptom (fraud) rather than the cause (AI music), but it’s easier to implement.

Policy updates. Spotify now requires distributors to disclose AI use and has banned unauthorized voice clones. These policies are new and still evolving.

What none of them are doing yet: fundamentally rethinking how streaming royalties work when anyone can generate unlimited content.

The current system pays based on streams. If you can generate 1,000 tracks in an hour and use bots to stream them, you can siphon money from artists who spent months on a single album. The detection systems help, but they’re playing defense against an offense that keeps getting more sophisticated.

What This Means for You

If you’re a creator using AI tools, the landscape is shifting fast.

The platforms are building detection systems. Your AI-assisted music may get flagged, even if you’re using it legitimately. The line between “AI-assisted” and “fully AI-generated” matters for how platforms treat your content.

The royalty pool is being diluted. Even with fraud detection, the sheer volume of AI content affects how money flows. More tracks competing for the same pool means less per stream for everyone.

The legal questions aren’t settled. Who owns AI-generated music? Can it be copyrighted? The Copyright Office says fully AI-generated works don’t get protection. Courts haven’t fully weighed in. Labels are still suing Suno and Udio.

The stigma is real. 40% of streaming users say they’d skip AI music without listening if they knew it was AI. 45% want the option to filter it out entirely. That’s not a majority, but it’s not nothing.

The Bigger Picture

The music industry has always adapted to new technology. Every format change, every distribution shift, every new tool has been met with predictions of doom followed by eventual adaptation.

AI is different in one crucial way: it doesn’t just change how music is made or distributed. It changes who can make it and how much can be created.

When 60,000 AI tracks hit a single platform every day, when 85% of those streams are fraudulent, when 2 billion fake streams get demonetized in a single year, we’re not talking about a creative tool. We’re talking about an industrial-scale challenge to the economics of music.

The 97% who couldn’t tell the difference in a listening test? They’re not the story.

The story is what we do about a system where anyone can create unlimited content, upload it instantly, and compete for the same royalty pool as artists who spend their lives making music.

That’s the conversation we should be having.

What to Watch

Detection technology licensing. Deezer is now selling their AI detection tool to other companies. If Spotify and Apple Music adopt similar systems, AI content will be flagged across all major platforms.

Royalty model changes. Deezer has already moved to an “artist-centric” payment model. Others may follow, shifting away from pure stream counts.

Legal developments. Sony is still suing Suno and Udio. The UK is debating copyright changes for AI training. The Copyright Office continues to clarify what can and can’t be protected.

Platform policies. Watch for changes in how distributors handle AI content, whether platforms require disclosure, and how recommendation algorithms treat synthetic music.

This is exactly the kind of complexity I’m tracking at ClearVerse. We’re building tools to help creators understand not just whether something sounds similar, but whether it creates actual legal risk. Because in a world where AI can make anything sound real, understanding the legal and economic reality becomes even more important.

If you’re navigating this, I’d love to hear what questions you have. What’s confusing? What’s worrying you? What do you wish the platforms would do?

Hit reply and tell me what you’re seeing.

We’re opening early access soon. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

— Christian

P.S. — Know someone trying to make sense of AI music? Send this their way.

Sinclair, a former Warner Music executive who joined Suno last July, was taking direct aim at Universal Music Group’s approach to AI licensing. His argument: locking AI-generated music inside closed platforms will stifle creativity and repeat the mistakes the industry made fighting the internet in the early 2000s.

Universal’s response came through Michael Nash, the company’s Chief Digital Officer, who called Suno’s open approach “an unsustainable status quo” that risks “direct cannibalisation” of artists.

This isn’t corporate posturing. This is a fundamental disagreement about how AI music should work. And whichever side wins will shape what you can and can’t do with AI music tools for years to come.

What’s Actually Being Debated

The fight comes down to one question: what happens after you create something with AI?

The walled garden model (Universal’s position): AI-generated music stays on the platform where it was created. You can stream it, share it within the app, maybe remix it further. But you can’t download it. You can’t put it on Spotify. You can’t use it in a YouTube video. The music exists only inside the walls.

The open studio model (Suno’s position): You can create music with AI and then take it wherever you want. Download it. Distribute it. Monetize it. The platform is a tool, not a container.

These aren’t theoretical positions. They’re already being implemented.

What Happened to Udio

When Universal settled its lawsuit with Udio in October 2025, the “walled garden” concept went from idea to reality overnight.

The day the deal was announced, Udio disabled all downloads. No warning. Users who had paid for subscriptions, who had spent months building libraries of AI-generated tracks, suddenly couldn’t access their own creations.

The backlash was immediate. Reddit threads filled with users threatening legal action. Some called it fraud. One user wrote: “I’ve spent hundreds of dollars and countless hours building tracks with this tool. No one warned us that one day, we wouldn’t even be able to access our own music.”

Udio offered a 48-hour window for users to download their existing tracks. After that, the walls went up permanently.

The new Udio platform launching later this year will operate entirely as a walled garden. You’ll be able to stream your creations within the app. You won’t be able to take them anywhere else.

Why Suno Got a Different Deal

A month after Udio’s settlement, Warner Music struck a deal with Suno. But the terms were notably different.

Suno users can still create songs. They can still download them. They can still distribute them to streaming services.

The main changes: Suno will launch new models trained on licensed music in 2026. Free users won’t be able to download (only stream and share). Paid users will have monthly download caps instead of unlimited access.

It’s a restriction, but it’s not a prison.

Why did Warner give Suno more freedom than Universal gave Udio? Industry observers point to a few factors.

First, Suno is bigger. With nearly 100 million users and a recent $250 million funding round valuing the company at $2.45 billion, Suno had more leverage.

Second, there’s the China question. Multiple analysts have noted that if Western AI music platforms become too restrictive, users will migrate to unregulated alternatives. Reddit threads following the settlements were full of comments warning: “Soon there will be a model from China to fill the gap.”

Third, Warner may simply have a different philosophy. CEO Robert Kyncl has been more publicly open to AI innovation than his counterparts at Universal and Sony.

The Arguments on Each Side

Universal’s case for walled gardens:

The labels argue that open distribution of AI-generated music will flood streaming platforms with low-quality content, dilute royalty pools, and directly compete with human artists. If anyone can generate a song and upload it to Spotify, the value of human-created music decreases.

They also point to fraud. Deezer reported in November 2025 that up to 70% of streams on AI-generated tracks are fraudulent, driven by bot farms gaming the royalty system. Walled gardens, they argue, prevent AI music from escaping into environments where it can be weaponized against artists.

Michael Nash put it bluntly: letting AI music flow freely is “direct cannibalisation” of artists’ work.

Suno’s case for open studios:

Sinclair argues that excessive restrictions will kill the creative potential of AI tools. In his LinkedIn post, he wrote: “If we had tried to lock music into closed systems over the last 25 years, we wouldn’t have streaming as we know it — the world music library, available in your pocket.”

His point: user-generated content and open platforms enabled the rise of global genres and viral discovery. SoundCloud launched careers. YouTube made music accessible to billions. Restricting AI music to walled gardens treats every user as a potential bad actor rather than a potential creator.

Sinclair also argues that openness is compatible with licensing. Suno’s Warner deal proves you can have licensed models and still let users export their creations.

What This Means for You

If you’re using AI music tools today, here’s the practical impact:

If you’re on Udio: Your options are shrinking. Once the new platform launches, you’ll be able to create music within the app, but you won’t be able to take it anywhere. If your goal is to distribute AI-generated music on streaming platforms, Udio is no longer the tool for that.

If you’re on Suno: You’ll still be able to create and download, but with limits. Free tier users will lose download access when the new models launch in 2026. Paid users will have caps. The open studio model survives, but it’s less open than before.

If you’re waiting to see who wins: Watch the Sony settlements. Sony hasn’t settled with either Suno or Udio. They’re still actively litigating. How Sony’s lawsuits resolve will signal whether the industry is moving toward walled gardens as the standard or accepting some version of open distribution.

The Bigger Picture

This debate isn’t just about AI music platforms. It’s about the future of AI-assisted creativity across every medium.

The walled garden model is a bet that control is more valuable than distribution. It prioritizes protecting existing revenue streams over enabling new ones. It assumes that if you give people tools to create and distribute freely, they’ll mostly use those tools to harm artists.

The open studio model is a bet that creativity flourishes when you remove friction. It prioritizes reach and accessibility over control. It assumes that most people using AI tools are genuine creators, not fraudsters, and that treating them like criminals will just push them to unregulated alternatives.

Both sides claim to be protecting artists. But they have fundamentally different ideas about what protection means.

What I’m Watching

Three things will determine how this plays out:

1. The quality gap. If the new licensed models from Suno and Udio sound noticeably worse than the current “trained on everything” versions, users will revolt or migrate to unregulated alternatives. The labels are betting that licensed training data can still produce compelling music. We’ll see.

2. International competition. If Chinese or other international AI music platforms offer unrestricted creation and distribution, users who care about flexibility will leave. The walled garden only works if there’s nowhere else to go.

3. Creator backlash or adoption. Will professional musicians embrace these tools as creative aids, or reject them as threats? Right now, the music industry is mostly fighting AI. But if major artists start publicly using AI-assisted creation, the dynamic shifts.

What I’m Building

I’ve spent the last six months building tools to help creators navigate the legal complexity of AI music. ClearVerse focuses on copyright risk assessment: understanding whether your music might trigger infringement claims before you publish.

But the walled garden debate adds a new layer. It’s not just about whether your music might infringe on someone else’s rights. It’s about whether you’ll have any rights at all to music you create using these tools.

We’re opening early access soon. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

One Last Thing

The music industry has a long history of fighting new technology, then adapting, then profiting from it.

They fought cassette tapes. They fought Napster. They fought YouTube. Each time, the initial response was restriction and litigation. Each time, the eventual outcome was licensing deals that created new revenue streams.

AI music might follow the same pattern. The question is whether the industry learned from those fights or is doomed to repeat them.

What do you think? Are walled gardens necessary protection or counterproductive overreach?

Hit reply and let me know. I read every response.

Talk soon.

— Christian

P.S. — If you know someone trying to figure out what these platform changes mean for their music, forward this their way.

Ghostwriter had used AI to clone their voices. The track was so convincing that listeners couldn’t tell the difference. Universal Music Group scrambled to get it taken down, and eventually succeeded. But the takedown wasn’t based on a clear legal right to the artists’ voices. It was a copyright claim over a producer tag buried in the track.

That workaround worked for Drake. But what about everyone else?

If you’re a musician, voice actor, podcaster, or anyone whose voice has value, here’s the uncomfortable truth: in most of the United States, someone can clone your voice with AI and there’s not much you can do about it.

The Patchwork Problem

The right to control how your voice and likeness are used commercially is called the “right of publicity.” It’s the legal principle that says you, not someone else, get to decide if your face appears on a billboard or your voice sells a product.

The problem is that right of publicity laws vary wildly from state to state. Some states have strong protections. Some have weak ones. And most of them were written long before anyone imagined you could generate a convincing vocal clone from a few minutes of audio samples.

Here’s what the landscape looks like:

Strong protection (explicit AI coverage): Tennessee passed the ELVIS Act in March 2024, which explicitly protects against AI voice cloning. It’s the first law in the country specifically designed for this. Violations can result in civil lawsuits and criminal prosecution as a Class A misdemeanor.

Moderate protection (voice included, but no AI-specific language): California has protected voice as part of publicity rights since the 1980s, thanks to cases like Bette Midler v. Ford Motor Company. But the law wasn’t written with AI in mind, so there’s uncertainty about how it applies.

Weak or no protection: Many states don’t explicitly include voice in their publicity rights at all. And even in states that do, most laws require the unauthorized use to be “commercial” in a traditional sense. AI-generated content doesn’t always fit neatly into that box.

The result is a patchwork where your rights depend entirely on where you live and where the person cloning your voice operates. If someone in a state with no voice protection clones your vocals and uploads the result, your legal options might be limited to nothing.

What Tennessee’s ELVIS Act Actually Does

Tennessee’s Ensuring Likeness Voice and Image Security Act (yes, named after Elvis) took effect on July 1, 2024. It’s worth understanding because it’s likely the template for future legislation elsewhere.

The ELVIS Act does a few key things:

Explicitly includes voice as a property right. The law defines “voice” broadly to include both an individual’s actual voice and any simulation of it. If it’s “readily identifiable and attributable” to you, it’s protected.

Covers AI-generated replicas specifically. The law targets “personalized generative AI cloning models and services that enable human impersonation.”

Creates liability for the tools, not just the content. You can sue not only the person who publishes an unauthorized voice clone, but also anyone who “makes available an algorithm, software, tool, or other technology” whose primary purpose is creating unauthorized voice replicas.

Carries real penalties. Civil lawsuits can seek injunctions, actual damages, and profits. Criminal violations are Class A misdemeanors, carrying up to 11 months and 29 days in jail and fines up to $2,500.

Has exemptions for legitimate use. News coverage, commentary, criticism, satire, and parody are protected. So is representing yourself as yourself in an audiovisual work, unless the work is designed to create a false impression that it’s authentic.

The music industry lobbied hard for this law. Luke Bryan, Chris Janson, Lindsay Ell, and other Nashville artists showed up at the signing ceremony. The bill passed the Tennessee legislature unanimously.

But Tennessee is just one state. And the AI tools that can clone your voice don’t respect state lines.

The Federal Void

There’s no federal right of publicity in the United States. That’s why you end up with 50 different state laws (and some states with no law at all).

Congress is trying to change that. The NO FAKES Act (Nurture Originals, Foster Art, and Keep Entertainment Safe) was first introduced as a draft in 2023, formally introduced in 2024, and reintroduced in April 2025. It would create a federal intellectual property right to your voice and likeness for the first time.

The bill has bipartisan support and backing from an unusual coalition: the music industry (UMG, Warner, RIAA, SAG-AFTRA) and the tech industry (YouTube, OpenAI, Google, Amazon, IBM) are both on board. That kind of alignment is rare.

What the NO FAKES Act would do:

• Establish that every individual has a federal property right in their voice and visual likeness

• Allow individuals (or their heirs, for 70 years after death) to control the use of digital replicas

• Create liability for creating, distributing, or profiting from unauthorized AI-generated replicas

• Set up a notice-and-takedown process similar to the DMCA for copyright

• Include exemptions for news, commentary, parody, satire, and other First Amendment-protected speech

If passed, it would replace the state-by-state patchwork with a uniform national standard.

But the bill hasn’t passed yet. It failed to advance in 2024. It was reintroduced in April 2025 with more industry support, but as of now, there’s still no federal protection. And even if it does pass, the takedown mechanisms and enforcement details will matter a lot for how effective it actually is.

What This Means for Different Creators

If you’re a professional musician or voice actor:

Your voice is your primary asset. Until federal law catches up, your protection depends heavily on where you’re based. If you’re in Tennessee, you have explicit protection against AI cloning. If you’re in California, you have some protection, but it wasn’t written for AI. If you’re elsewhere, check your state’s publicity rights laws carefully.

Consider whether your contracts with labels, publishers, or platforms address AI voice cloning. Many older contracts don’t. Newer ones increasingly do, often with opt-in provisions for AI training or likeness licensing.

If you’re an independent creator building an audience:

You’re probably not thinking about voice cloning as a threat yet. But as your profile grows, so does the value of your voice. Someone could theoretically train an AI on your podcast episodes, YouTube videos, or social content and create synthetic versions of you.

Documentation matters. Keep records of your original content. If you ever need to prove that your voice was cloned without permission, having timestamps and originals helps.

If you’re using AI voice tools:

Be careful whose voice you’re cloning. Using AI to generate content that sounds like a real, identifiable person without their permission could expose you to liability, especially if you’re in Tennessee or if the cloned artist is.

The “Heart on My Sleeve” case showed how fast platforms will act when major labels are involved. But smaller artists have less leverage. That doesn’t mean the legal risk is zero.

The Practical Playbook

Until federal law provides clear protection, here’s what you can do:

1. Know your state’s law

Look up whether your state has a right of publicity statute, whether it includes voice, and whether it applies to AI-generated content. Tennessee’s ELVIS Act is the gold standard. California offers moderate protection. Many states offer very little.

2. Document your voice

Keep records of your original vocal work with timestamps. If you ever need to prove your voice was cloned, this documentation will matter.

3. Review your contracts

If you’re signing with labels, publishers, or platforms, look for clauses about AI, voice likeness, and digital replicas. Some newer deals include opt-in provisions for AI training. Understand what you’re agreeing to.

4. Monitor for unauthorized use

This is harder than it sounds. Unlike visual content, there’s no reverse-image search for voice. But keeping an eye on social media, YouTube, and streaming platforms for content falsely attributed to you is increasingly important.

5. Support federal legislation

The NO FAKES Act would solve a lot of these problems. If it matters to you, let your representatives know.

What I’m Building

I’ve spent the last six months deep in copyright and AI law for music. The voice cloning issue sits right at the intersection of what ClearVerse is about: helping creators understand their legal exposure before it becomes a problem.

Right now, ClearVerse is focused on copyright risk assessment. But the broader question is the same: what are your actual rights, and how do you protect them in a landscape that’s changing faster than the law can keep up?

We’re opening early access soon. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

One Last Thing

The Ghostwriter incident was a wake-up call. But it won’t be the last time someone clones a famous voice and goes viral. The tools are getting better. The barrier to entry is getting lower. And the legal framework still has massive gaps.

I want to hear from you: Has anyone ever used your voice or likeness without permission? Would you even know if they did?

Hit reply and let me know. I read every response.

Talk soon.

— Christian

P.S. — If you know a creator who hasn’t thought about voice cloning, send this their way. This is one of those things where the first time you learn about it shouldn’t be when it happens to you.

The message is clear: if a machine made it, you don’t own it.

This might sound like an abstract legal debate. It’s not. If you’re using Suno, Udio, or any AI tool to make music, this affects you directly. Because that track you spent an hour prompting and tweaking? It might belong to everyone. And anyone can copy it, remix it, or upload it as their own.

Here’s what’s actually going on.

The Rule: Human Authorship Required

In January 2025, the U.S. Copyright Office released its official guidance on AI-generated works. The core finding: copyright protection requires human authorship. If an AI determines the expressive elements of a work, that work isn’t copyrightable.

The key phrase is “meaningful human authorship.” The Copyright Office was explicit: prompts alone don’t count.

Here’s how they put it: “Given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output.”

So if you type “make me a lo-fi hip hop beat with jazzy piano samples” into Suno and download the result, you don’t own it. Not legally. That track enters the public domain the moment it’s created.

This isn’t a technicality. It’s the foundation of copyright law. And courts are enforcing it.

What This Means in Practice

Let’s say you create a song with Suno. You love it. You upload it to Spotify. It starts getting streams.

Then someone else downloads your track, uploads it under their name, and starts collecting your royalties.

What can you do about it?

If your track was purely AI-generated, the answer might be: nothing.

Copyright gives you the legal right to control who copies, distributes, and profits from your work. But that right only exists if you own the copyright in the first place. If your work isn’t copyrightable because it lacks human authorship, you have no legal claim. Anyone can use it however they want.

This isn’t hypothetical. Streaming platforms are already flooded with AI-generated content. Deezer reported receiving over 50,000 AI-generated tracks per day in late 2025. As this flood continues, disputes over who “owns” AI-generated music will only increase.

And if you can’t prove meaningful human authorship, you’ll lose those disputes before they even start.

What Counts as “Meaningful Human Authorship”?

This is where it gets complicated. The Copyright Office evaluates authorship on a case-by-case basis. But their guidance gives us some clear signals about what does and doesn’t count.

What probably doesn’t count:

• Typing a text prompt into an AI generator

• Selecting from multiple AI-generated outputs

• Making minor tweaks to AI-generated content

• Choosing parameters like tempo, genre, or mood before generation

What probably does count:

• Writing original lyrics that you then have AI set to music

• Composing a melody yourself and using AI for arrangement or production

• Substantially editing, rearranging, or modifying AI-generated output

• Playing instruments over AI-generated stems

• Creating a “selection and arrangement” of AI elements that reflects your creative vision

The distinction the Copyright Office keeps making: are you using AI as a tool to assist your creativity, or is AI standing in for human creativity entirely?

If you’re a songwriter who uses AI to generate backing tracks for your original lyrics and melodies, you probably have a stronger copyright claim than someone who prompts an AI to create an entire song from scratch.

The Spectrum of Ownership

Think of AI music creation on a spectrum:

Fully AI-generated (no copyright): You prompt Suno with “upbeat pop song about summer love” and download the result without modification. The AI determined the melody, harmony, rhythm, lyrics, and arrangement. You contributed a concept. That’s not enough.

AI-assisted with minimal human input (probably no copyright): You generate 10 versions and pick your favorite. You trim the intro and adjust the volume. You added selection and minor editing, but the expressive elements were still determined by the AI.

AI-assisted with substantial human input (probably copyrightable): You write original lyrics. You hum a melody into the AI and have it develop the arrangement. You export the stems and re-record the bass line yourself. You make significant creative decisions that shape the final work. The human authorship is demonstrable.

Human-created with AI assistance (copyrightable): You compose the song yourself, play the instruments, write the lyrics. You use AI for mixing assistance or to clean up a vocal take. The AI is a tool, like Auto-Tune or a drum machine. The creative expression is yours.

The further right you are on this spectrum, the stronger your copyright claim.

Why the Thaler Case Matters

Stephen Thaler has been fighting this battle for years. He created an AI system called the “Creativity Machine” and tried to register a work it generated as copyrighted. He listed the AI as the author and himself as the owner.

The Copyright Office rejected him. A federal district court upheld that rejection. The D.C. Circuit Court of Appeals affirmed it in March 2025. And in May 2025, the court declined to rehear the case.

Now Thaler has petitioned the Supreme Court. His argument: the Copyright Act doesn’t explicitly require human authorship, and the Copyright Office is adding words to the statute that Congress never wrote.

On January 23, 2026, the Department of Justice filed its response urging the Court not to take the case. Their position: the lower courts got it right, and this case doesn’t raise any new legal questions that need Supreme Court resolution.

If the Supreme Court declines to hear the case (which seems likely), the human authorship requirement stays in place. And every creator using AI tools needs to understand what that means for their work.

The Practical Problem: Documentation

Even if you do add meaningful human authorship to your AI-assisted work, you might have trouble proving it.

The Copyright Office requires applicants to disclose AI-generated content and describe their human contributions. If you can’t articulate what you contributed, your registration could be rejected or challenged later.

This creates a documentation problem. Most creators aren’t thinking about copyright when they’re in the middle of making music. They’re not logging which elements they wrote versus which the AI generated. They’re not saving drafts that show their creative process.

But if ownership ever becomes disputed, that documentation could be the difference between keeping your rights and losing them.

What To Do About This

If you’re using AI to make music and you want to actually own what you create, here’s the practical playbook:

1. Add substantial human creativity

Don’t just prompt and download. Write your own lyrics. Compose your own melodies. Re-record elements. Make creative decisions that shape the final work in demonstrable ways.

2. Document your process

Keep records of what you contributed versus what the AI generated. Save intermediate drafts. Note which elements you wrote, performed, or substantially modified. This documentation could matter if you ever need to prove authorship.

3. Understand your platform’s terms

Suno and Udio both have terms of service that grant you certain rights to your creations. But platform terms don’t override copyright law. If your work isn’t copyrightable, the platform can’t make it copyrightable. Read the fine print and understand what you’re actually getting.

4. Consider your use case

If you’re making music for fun and don’t care about ownership, none of this matters. Generate away. But if you’re planning to monetize, license, or commercially exploit your AI-assisted music, the copyright question becomes critical.

5. Stay informed

This area of law is evolving rapidly. The Copyright Office is still developing its guidance. Courts are still deciding cases. What’s true today might shift as the legal landscape develops.

The Bigger Picture

The Copyright Office’s position isn’t anti-AI. They’ve explicitly said that using AI as a tool doesn’t disqualify your work from copyright protection. The question is whether you’re the author or the AI is.

This distinction actually makes sense. Copyright exists to incentivize human creativity. It gives creators exclusive rights to their work so they can benefit from it. If machines can generate unlimited content without human creative input, extending copyright to that content doesn’t serve the original purpose.

But the practical implications for creators are real. If you’re building a catalog of AI-generated music, you need to understand that catalog might not be legally yours to protect.

What I’m Building

I’ve spent the last six months deep in copyright law and AI music. The more I learn, the more I see how much confusion exists between what platforms do, what the law says, and what creators assume.

ClearVerse is building tools to help creators navigate this landscape. Right now, we’re focused on copyright risk assessment: helping you understand whether your music might trigger infringement claims before you publish. But the ownership question is just as important. If you can’t own your work, protecting it becomes a different problem entirely.

We’re opening early access soon. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

One Last Thing

The AI copyright question isn’t going away. The Thaler case could get resolved this year. The Copyright Office is still releasing guidance. And millions of creators are generating music without understanding who actually owns it.

I want to hear from you: Have you thought about this before? Does it change how you’ll use AI tools?

Hit reply and let me know. I read every response.

Talk soon.

— Christian

P.S. — If you know someone making music with AI who hasn’t thought about ownership, send this their way.

The headlines say “landmark deals” and “victory for the creative community.” But when you read the actual details, the picture gets complicated. Some things changed. Some things didn’t. And some questions that actually matter to creators using these tools are still unanswered.

Here’s what’s really going on.

What Actually Happened

In June 2024, all three major labels (Universal, Sony, Warner) sued both Suno and Udio for copyright infringement. The allegation: these companies trained their AI models on millions of copyrighted songs without permission.

Then, starting in late October 2025, the settlements started rolling in:

UMG + Udio (October 29): Settlement plus licensing deal for a new “walled garden” platform launching in 2026.

WMG + Udio (November 19): Similar deal. Opt-in licensing for artists.

WMG + Suno (November 25): Settlement plus partnership. Suno also acquired Songkick from Warner as part of the deal.

Still fighting: Sony hasn’t settled with either company. UMG’s lawsuit against Suno is still active. As of late December, Sony filed documents confirming it plans to keep litigating against Udio.

So the scorecard looks like this:

Universal: Settled with Udio ✓ | Still suing Suno
Warner: Settled with both ✓
Sony: Still suing both

The lawsuits aren’t over. The legal questions aren’t answered. Two out of three majors found it cheaper to make deals than to keep fighting. Sony apparently disagrees.

What Changes for Suno and Udio Users

Both platforms announced changes coming in 2026. Here’s what we know:

Suno (under the Warner deal):

• New “licensed models” trained only on authorized music (launching later in 2026)

• Current models will be phased out when new ones launch

• Free tier users will no longer be able to download audio (only stream and share)

• Paid users will get download caps with option to pay for more

• Artists can opt-in to allow their voice, name, likeness in AI generations

Udio (under the UMG and Warner deals):

• Pivoting from open generation to a “fan engagement” platform

• New platform launching in 2026, trained on licensed music

• Becomes a “walled garden” where creations can’t leave the platform

• Focus shifts to remixing, mashing up, and interacting with licensed music

The key difference: Suno gets to keep doing what it does (generate new music from prompts) but with licensed training data. Udio has to fundamentally change its product into something closer to an interactive streaming service.

Industry observers are calling Suno’s deal significantly better. Udio essentially agreed to stop competing in the generative music space.

What Doesn’t Change

The fair use question is still unresolved. No court has ruled on whether training AI on copyrighted music constitutes fair use. The settlements specifically avoided setting legal precedent. Sony’s ongoing lawsuits could still produce a ruling, but we’re likely looking at 2026 or later.

Your old creations are still in legal limbo. The settlements cover future models trained on licensed content. They don’t retroactively bless everything created on current models. If you made something on Suno or Udio in 2024 or 2025 that sounds suspiciously like a copyrighted song, that liability hasn’t disappeared.

Indie artists aren’t covered. These deals are between the major labels and the AI companies. Independent musicians whose work was allegedly used in training? They’re still suing. Class action lawsuits were filed in October 2025, and more arrived in December. The indie artist cases are separate from the major label settlements.

Your AI-generated outputs might still be uncopyrightable. Remember: the Copyright Office ruled in January 2025 that purely AI-generated works can’t be copyrighted. These licensing deals don’t change that. If you prompt Suno and download the result without meaningful human modification, you still don’t own it.

What This Means for Different Types of Creators

If you’re using Suno to make finished songs:

You can keep doing this, but the terms are changing. Once the new licensed models launch later in 2026, free tier users won’t be able to download at all (only stream and share), and paid users will face download caps. Until then, the current models are still operating during the transition period. Your future creations will be trained on licensed music, which reduces (but doesn’t eliminate) infringement risk. But you still probably can’t copyright the output unless you add substantial human input.

If you’re using Udio:

The product is changing fundamentally. By 2026, Udio becomes a fan engagement platform, not a generative tool. You won’t be creating new songs from scratch. You’ll be interacting with licensed music in a controlled environment. If that’s not what you signed up for, look elsewhere.

If you’re an independent artist worried about your music being used:

The major label settlements don’t help you. Your best options are (1) join or follow the indie class action lawsuits, (2) opt-out of AI training through services like Spawning.ai, or (3) document your catalog meticulously in case you need to prove infringement later.

If you’re monetizing AI-generated music:

Be careful. Download restrictions and walled gardens suggest the labels want to control where this content ends up. Deezer estimates 50,000 AI-generated songs hit streaming services daily. The majors are clearly trying to stop that flood. Expect distribution platforms to tighten their policies.

The Uncomfortable Questions Nobody’s Answering

Where does the money go?

The deals promise “compensation” for artists and songwriters, but nobody’s disclosed the actual terms. Irving Azoff, legendary artist manager and founder of the Music Artists Coalition, called it out immediately: “We’ve seen this before — everyone talks about ‘partnership,’ but artists end up on the sidelines with scraps.”

The announcements emphasize opt-in. But opt-in to what, exactly? What’s the royalty rate? How is usage tracked? How do you even know if your work is being used? None of this has been disclosed.

What about the music already created?

Suno has over 100 million users. Millions of songs have been generated. The settlements say current models will be “deprecated” in 2026. But those creations don’t disappear. What’s their legal status? The settlements are silent.

Why did Warner and Universal settle but Sony didn’t?

Sony filed documents in December 2025 confirming it’s continuing litigation against both companies. That’s interesting. Either Sony thinks it can get a better deal, or it genuinely wants a court ruling on fair use. If Sony wins, it could reshape the entire AI music landscape. If it loses, the labels who settled early might look smart.

Are these deals actually enforceable?

Training AI on “only licensed music” sounds clean. But how do you verify it? The original lawsuits alleged Suno and Udio scraped songs from YouTube using stream-ripping techniques. How do the labels know the 2026 models won’t still have that data baked in? Fingerprinting and filtering were mentioned. Whether that’s sufficient remains to be seen.

What I Think Is Really Happening

Reading between the lines, here’s my take:

The major labels realized litigation was expensive and uncertain. Fair use cases are unpredictable. Meanwhile, Suno just raised $250 million at a $2.45 billion valuation. These AI companies aren’t going away.

So the labels did what labels do: they found a way to get paid. Licensing deals mean revenue. Partnerships mean influence over how the platforms develop. Walled gardens mean the output doesn’t compete directly with their catalogs on Spotify.

Is this good for artists? Maybe. The opt-in model gives creators some control. But the financial terms are opaque, and history suggests the middlemen (labels, platforms) capture most of the value.

Is this good for users? Mixed. Suno keeps working. Udio becomes something different. Both get more expensive and more restricted.

Is this good for the broader legal question of AI and copyright? No. These settlements kicked the can down the road. We still don’t know if training on copyrighted music is fair use. We still don’t know how courts will handle AI-generated outputs that resemble copyrighted works. The only way we get answers is if Sony pushes its cases to a ruling, or if the indie artist class actions succeed.

What to Watch in 2026

Here’s what I’m tracking:

Sony’s litigation. If they push to a ruling, we finally get legal clarity. If they settle too, the fair use question stays open indefinitely.

The indie artist class actions. These cases are specifically about whether AI-generated music creates “competitive, radio-quality substitutes” that harm the market for human-made music. That’s a different argument than the major label cases, and it could produce different results.

How opt-in actually works. When Warner and Universal artists start seeing (or not seeing) AI-related payments on their statements, we’ll know whether “compensation” is real or marketing.

Distribution platform policies. Spotify, Apple Music, and others will have to decide whether to accept music made on these platforms. Expect tightening.

The Copyright Office. Part 3 of their AI report came out in May 2025 and questioned whether AI training on music qualifies as fair use. More guidance could be coming.

What I’m Building

This is exactly why ClearVerse exists.

The settlements don’t solve the fundamental problem: creators have no way to assess their actual legal risk. Not the platform’s risk. Not the label’s risk. Yours.

You need to know: Does this track I made sound enough like a copyrighted song to get me sued? Not flagged by Content ID. Sued.

That’s the gap we’re filling. Technology that scores copyright risk based on how courts actually rule, not just how similar something sounds.

We’re opening early access soon. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics.

One Last Thing

If you’re using Suno or Udio, I want to hear from you.

Are you worried about these changes? Do you understand what you can and can’t do with your creations? Are you getting any clarity from the platforms themselves?

Hit reply and tell me: What’s your biggest question about AI music right now?

I read every response. Next week I’m planning to tackle whatever comes up most.

Talk soon.

— Christian

P.S. — If this was useful, forward it to someone else trying to figure out what these deals actually mean.

Behind every one of those claims is a creator staring at a notification, trying to figure out: Am I in legal trouble? Should I dispute this? What actually happened?

Most of them will never get a straight answer. The systems designed to “protect” copyright don’t explain copyright. They just enforce policies that have almost nothing to do with the law.

I’ve spent the last six months deep in the weeds of music copyright, building technology that translates similarity detection into actual legal risk. The deeper I’ve gotten, the more I’ve realized that most of what creators believe about copyright is just... wrong. And AI is making it worse.

Here are the five myths I keep running into.

Myth #1: “Similar” Means “Stolen”

When Content ID flags your video, it’s not saying you committed copyright infringement. It’s saying something in your upload (audio, video, or even a melody) matches something in their database. That’s it.

Content ID doesn’t check whether you have a license. It doesn’t evaluate fair use. It doesn’t ask whether the similar portion is legally “substantial” or whether an average listener would even recognize the similarity.

Courts look at all of that. Platforms look at none of it.

Here’s a situation I see constantly: You license a drum loop from Splice. You use it legally. But 500 other creators also licensed that same loop. Content ID flags all of you against each other, because it can’t tell the difference between “this sounds similar” and “this is infringement.”

You get a claim on content you have every legal right to use. The platform treats you like an infringer. And you’re left wondering if you did something wrong.

You didn’t. The system just can’t tell the difference.

Myth #2: “Under 50% Similarity Is Safe”

I hear this one all the time. “I’m only at 30% similarity, so I should be fine.”

There is no percentage threshold in copyright law. Not 50%. Not 30%. Not 10%.

The “70-30 rule” and “6-second rule” you’ve heard about? Those are platform policies. Internal guidelines that YouTube, TikTok, and Instagram use to manage their systems. They have nothing to do with what would happen in a courtroom.

Courts care about different questions entirely.

Did you copy the “heart” of the work? Two seconds of a hook that everyone recognizes can be more legally significant than two minutes of generic background music. Would an average listener recognize your track as coming from the original? Did you copy something that’s actually protectable, or just common musical elements that nobody owns?

An AI tool can tell you “this is 47% similar.” It cannot tell you whether that 47% includes the one melodic phrase that would get you sued.

Here’s where AI makes this worse: generative music tools can create infinite variations that hover right around similarity thresholds. But “similar enough to avoid detection” is not the same as “legally safe.” You can have zero Content ID claims and massive legal exposure if you copied the wrong four bars.

Myth #3: “If I Can’t Hear It, It’s Not Infringement”

Your ears are not a legal instrument.

In the Blurred Lines case, the jury found infringement even though the songs don’t share a single identical melody line. The similarity was in the “feel,” the groove, the combination of elements that created a recognizable vibe. I’m still not entirely sure I agree with that verdict. A lot of people in the industry don’t. But it’s the precedent we’re working with.

The Dark Horse case went the other direction. A jury initially found that Katy Perry’s song infringed a Christian rap track based on an 8-note repeating pattern. But an appeals court overturned it, ruling the pattern was too common to be copyrightable. Same type of claim, opposite outcome.

What these cases show is that copyright analysis is multidimensional. Courts look at melody, harmony, rhythm, structure, timbre, lyrics. They don’t weight these equally. Melody is generally more protected than rhythm. A match in one dimension can be legally significant even if everything else is completely different.

You might listen to your track and think “this sounds nothing like the original.” But if you inadvertently copied a melodic contour or a harmonic progression, you might have legal exposure you literally can’t hear.

That’s not paranoia. It’s just how the law works. And no detection tool on the market explains which dimensions matter in your specific case.

Myth #4: “AI-Generated Music Is Safe to Use”

This one’s for creators using AI music tools, and it’s more complicated than most people realize.

The training question is still unresolved. In 2024, the major labels sued Suno and Udio for copyright infringement, alleging that training on copyrighted music without permission was itself illegal. Since then, some cases have settled. Warner Music Group is now partnering with Suno. Universal settled with Udio. But Sony’s case against both companies is still active, and no court has definitively ruled on whether training AI on copyrighted music constitutes fair use. The settlements avoided setting precedent. The legal question remains open.

But here’s what most creators don’t realize: Even if AI training is eventually ruled legal, your AI-generated output might not be copyrightable at all.

In January 2025, the U.S. Copyright Office ruled that purely AI-generated works cannot receive copyright protection. A federal appeals court affirmed this in March. The logic: copyright exists to protect human creativity, and works generated entirely by AI lack the required human authorship.

What does this mean practically? Music you create by typing a prompt into Suno or Udio and hitting generate? That’s public domain. Anyone can use it, remix it, or monetize it. You have no legal protection.

The only way to own what you create is to add meaningful human involvement. Edit the melodies. Write your own lyrics. Add live instrumentation. Layer your own production. The Copyright Office has said AI-assisted works can qualify for protection if the human contribution is substantial enough, but you’ll need to document your creative process if you ever want to register or defend it.

So when creators assume that because an AI tool exists, the output must be legal to use and theirs to own, that’s not how any of this works.

Pay attention to how these cases develop. Don’t assume “AI-generated” means “copyright-free.” And if you want to actually own your music, make sure you’re doing more than prompting.

Myth #5: “Copyright Detection Tools Know the Law”

This is the myth that made me start building ClearVerse.

Every major detection company (Audible Magic, Pex, YouTube’s Content ID) does the same thing: measure similarity. They compare audio fingerprints, identify matches, and return a number.

That number tells you nothing about legal risk.

This one surprised me when I found it: 65% of Content ID disputes succeed. Most creators don’t dispute because they’re scared of getting a strike. But the majority of claims, when actually challenged, get resolved in the creator’s favor.

Why? Because detection and law are measuring different things.

Detection asks: Does this sound similar?

Law asks: Is this similarity the kind that matters?

A detection tool can’t tell you whether the similar portion is legally “substantial.” It can’t evaluate fair use. It can’t tell you if the element you matched is even copyrightable, or what a court would actually do with your case.

What creators need isn’t just detection. It’s interpretation. Not just “47% similar,” but “here’s what that similarity means given how courts have ruled on cases like this.”

That gap is why I started building what I’m building.

What This All Means

The copyright system wasn’t designed for a world where anyone can generate infinite musical variations with AI, where platforms process billions of claims automatically, where detection technology is completely divorced from legal analysis, and where creators bear all the risk without any tools to assess it.

The result is a system that optimizes for one thing: don’t get the platform sued.

That’s rational from YouTube’s perspective. It’s terrible for creators.

You’re navigating a minefield where claims don’t mean you broke the law, similarity scores don’t predict legal outcomes, platforms have no incentive to explain the difference, and the only people who really understand the nuances charge hundreds of dollars an hour.

What I’m Building

I started ClearVerse because I kept seeing the same pattern: sophisticated detection, zero legal context.

We’re building something different. Technology that scores copyright risk based on how courts actually rule, not just how similar something sounds.

And yes, ClearVerse uses AI. But there’s a difference between AI that generates content without understanding legal risk and AI that helps you understand risk before you create. The same technology that’s flooding the market with potential copyright landmines is also the only thing capable of analyzing legal patterns at scale. It comes down to what you point it at.

We’re using AI to do what humans can’t do fast enough: cross-reference your content against how courts have actually ruled on similarity, weigh which musical elements carry legal weight, and translate that into a risk score you can act on.

The goal is to give creators the intelligence they need to make informed decisions before they upload, not after a claim hits their channel.

We’re opening early access soon. Sign up at clearverse.ai to be first in line.

Want a walkthrough? If you’re a label, agency, or high-volume creator, I’d love to show you what we’re building. Just reply to this email or DM me on LinkedIn.

Or just subscribe. I’ll be writing more about AI, copyright, and creator economics. Next week I’m looking at what the Suno and Udio settlements actually mean for creators, and what’s still unresolved.

One Last Thing

If you’ve gotten this far, you probably have opinions. Maybe you’ve dealt with bogus claims. Maybe you’ve lost revenue to Content ID. Maybe you think I got something wrong.

I want to hear it.

Hit reply and tell me: What’s your biggest copyright frustration right now?

I read every response. Some of the best article ideas come from questions I hadn’t thought to ask.

Talk soon.

— Christian

P.S. — Know a creator dealing with copyright headaches? Send this their way.

Subscribe now