All articles

Suno Sued Again: What the Second Universal and Sony Lawsuit Means for Your Music

By

Published Updated

Two days ago, on Friday, September 18, Universal Music Group and Sony Music Entertainment sued Suno for the second time. I read the coverage that evening with the same knot in my stomach I had back in June 2024, because I use this tool, half the producers I know use this tool, and plenty of people reading this have released songs that started life as a Suno generation.

Here is the short version before we go deep. The new complaint, filed in Boston federal court, accuses Suno of copying 60,202 of the labels’ sound recordings without a license, and it claims the brand new v6 models, the ones built with Warner, BMG, and Believe, are still contaminated by everything that came before. Nobody is suing you. Your released songs are not named anywhere in this Suno lawsuit. But the platform you are standing on has been rearranging itself under your feet all year, and this filing tells you which way it will keep moving.

So let’s walk through what actually happened, what it changes for working musicians, and what I would do about it this week.

The Filing in Plain English

The complaint landed on Friday in the US District Court for the District of Massachusetts. Twelve label entities are listed as plaintiffs, led by UMG Recordings and Sony Music Entertainment, and according to Music Business Worldwide, which obtained the filing, it accuses Suno of copying 60,202 of their recordings to build the models behind its music generation service.

That number matters because of how US law calculates statutory damages, meaning fixed dollar amounts a court can award per infringed work without the labels having to prove actual losses. Willful infringement tops out at $150,000 per work, which puts the theoretical ceiling here just over $9 billion. The labels are also asking for up to $2,500 for each act of circumventing YouTube’s download protections, a separate claim under the DMCA, the Digital Millennium Copyright Act, which is the US law that bans breaking digital locks on copyrighted content. Applied once per recording, that adds roughly another $150 million.

On top of the money, Universal and Sony want a court order stopping Suno from any further use of their recordings, and they have demanded a jury trial.

Why a second lawsuit instead of a bigger first one? The original case, filed in June 2024, covers just 560 works, because at the time the labels could only guess at what sat inside Suno’s training data. During discovery they ran audio fingerprinting from Audible Magic against that data and came back with more than 61,000 matches. They asked Judge F. Dennis Saylor IV to fold those works into the existing case, and on August 18 he refused, ruling the additions would delay a case he wants decided on fair use, and noting the labels could simply bring the new works as a separate action. So they did.

One more piece of context from the first case. In an answer filed on September 1, Suno admitted it obtained audio from YouTube using YT-DLP, a free downloading tool, while disputing whether the labels even have standing to bring that stream-ripping claim. Fact discovery in that case closed on September 30, so the evidence about Suno’s training data is now locked into the record and the fair use fight is coming to a head.

Why v6 Is the Real Target

Here is what makes this second complaint genuinely new, and honestly a little chilling if you build anything with AI. On September 9, Suno launched its v6 model family in partnership with Warner Music Group, BMG, and Believe, and retired every earlier model the same day. Chief Product Officer Jack Brody told Music Business Worldwide that “v6 was trained entirely from scratch, from the ground up,” on data that does not include Universal or Sony recordings.

The labels’ answer, nine days later, is that it does not matter. Their complaint argues v6 was trained on user interactions with the older models. Every time you generate on Suno, it gives you two versions of a song and remembers which one you picked, and that preference data, along with the audio outputs themselves, fed the new models.

The filing also alleges Suno used knowledge distillation, where a new student model is trained to mimic the behavior of an older teacher model. Since every teacher in that chain learned from the unlicensed catalog, the labels argue the value of their recordings passed straight through into v6. In their words, v6 is not a fresh start but “the fruit of the same poisoned tree.”

The complaint names v4, v4.5, v4.5+, v4.5-all, v5, and v5.5 as trained on the copied corpus, and it points out that Suno has never claimed it destroyed the copies it made. Coverage of the filing by The Next Web adds that the labels also flagged how Suno quietly removed its old block on prompting with artist names.

Suno’s public position, as reported by Variety, has been that its training metadata does not include artist names, that its models cannot reproduce the material they trained on, and that training its early models was fair use. Fair use is the US doctrine that lets some unlicensed uses of copyrighted work stand, weighed on factors that include harm to the market for the originals.

I want to be straight about why this section matters beyond one platform. No US court has ruled on whether retraining a model on the outputs of an earlier, allegedly infringing model washes it clean. If the poisoned tree theory sticks, saying we retrained on licensed data stops being an escape hatch for the entire AI industry. If it fails, licensed retraining becomes the standard cleanup playbook. Either way, this case will get cited in AI music copyright fights for a decade.

The Other Suits Suno Is Fighting

The two label cases are not the only ones on the docket. On August 17, the independent publisher Round Hill Music sued Suno in the Northern District of California, alongside a separate suit against Anthropic over song lyrics used to train its chatbot. Round Hill says it may expand the case to ten thousand or more compositions, that damages could approach a billion dollars, and that it intends to go to trial rather than settle for a licensing deal. And on August 20, a federal judge in Massachusetts refused to dismiss most of a proposed class action brought by independent artists against Suno, letting claims about AI output and stream ripping move forward into discovery.

How Warner Got Out and Why Universal and Sony Stayed In

Warner sued Suno alongside Universal and Sony back in 2024. Then, in November 2025, Warner settled and struck the first licensing deal of its kind with the company, built on a promise of licensed models arriving in 2026. BMG signed on August 12 of this year, and Believe announced its partnership on September 8, with Believe and its distributor TuneCore stressing that artists’ music would only be included when each artist chooses to opt in.

Rival Udio went a different route. It settled Universal’s lawsuit in October 2025, switched off downloads entirely, and now holds licensing deals with Universal, Warner, Merlin, and Kobalt. Udio runs what the industry calls a walled garden, the model Universal has championed, where you make music on the platform and it stays on the platform. Suno has resisted that, kept downloads alive, and its Chief Music Officer Paul Sinclair argued publicly in February for open studios instead of walled gardens. Sony, for what it’s worth, is not letting Udio off either. It sued Udio a second time in July over 30,000 additional songs.

It is worth pausing on what that difference means for the person actually making songs. In a walled garden your music lives as content inside someone else’s app, streamable and shareable there but never a file on your hard drive, while an open studio hands you an actual audio file you can mix, register, and release like any other recording. Every practical step later in this article depends on having that file, which is why downloads keep sitting at the center of these settlements.

Here is the twist. Universal and Sony now use Suno’s own deals as a weapon. Three agreements with major rightsholders in under a year, the complaint argues, prove that a functioning market exists for licensing recordings to train AI models, and prove that Suno itself knows this use requires permission. That cuts directly at the market harm factor in Suno’s fair use defense. Brody has said the revenue sharing in those deals is not payment for training data, and the complaint essentially rolls its eyes at that framing.

The money in the background explains why nobody is walking away. In June, Bloomberg reported that Suno raised $400 million at a $5.4 billion valuation in a round led by Bond Capital, more than double the $2.45 billion it was worth after its November 2025 raise. Bloomberg’s reporting put the platform past 2 million paying subscribers earlier this year, with roughly $300 million in projected annual revenue. A company that size is worth suing, worth licensing to, and able to fund a long fight, and all three things are happening at once.

What Changed for Suno Users on September 3

While the lawyers were filing motions, the product you actually pay for changed underneath you. Suno published new terms on August 10 that took effect September 3, and Music Business Worldwide’s breakdown of the new download limits is worth reading in full, because these rules now govern every song in your library.

Pro subscribers, who pay $8 to $10 a month depending on whether they bill annually or monthly, are now capped at 20 song downloads per month. Premier subscribers, at $24 to $30 a month, get 60, though Premier users can still download without limit from Suno Studio, the browser-based production suite. Both tiers previously had unlimited downloads, so this is a real haircut, softened only by the option to buy extra downloads when you run out.

Free users got it rougher: seven downloads total, ever, and those files are cleared for personal use only. No streaming platforms, no selling, no client work. If you have been sitting on the free tier telling yourself you will upgrade once a song takes off, that plan quietly stopped working this month.

Two other rules deserve your full attention. First, a song in your library cannot be released, sold, or licensed until you have spent a download on it, so downloads now function as commercial clearance rather than simple file access. Second, pulling audio out any other way, by recording your screen or stream ripping your own track, is explicitly banned. Suno reserves the right to stamp outputs with a fingerprint, watermark, or metadata showing your tier and whether the file was a permitted download, stripping those markers is itself a violation, and enforcement can run from removal to account termination to a report to law enforcement.

The remix rules tightened as well. A remix of another user’s track is now cleared for personal use only, so you cannot take someone’s public generation, rework it, and push the result to streaming or a sales page. If you want to collaborate, do it the old-fashioned way, with stems, consent, and a split sheet, outside the platform.

Suno frames the caps as a way to stop bad actors from mass exporting music, and there is real history behind that. Earlier this year an artists’ rights coalition publicly accused the platform of feeding streaming fraud at industrial scale. There is a regulatory push behind the watermark language too: the EU AI Act requires machine-readable markers in AI-generated content, and services already on the market before August 2, 2026 have until December 2, 2026 to comply. So the marking is coming regardless of how any lawsuit goes.

My practical advice here is simple. Treat downloads like studio time and budget them, and when a track matters, spend the download and archive the file the day you finish it. Twenty a month sounds tight until you get selective, and the cap has honestly pushed me to finish songs instead of hoarding half-finished ideas. The subscribers these limits really hurt are running upload farms, not writing records.

Your Released Songs Are Not the Target, but Read the Fine Print

Let me lower the temperature on the question I have been asked most since Friday. No, the labels are not suing Suno users. The complaint targets how the models were trained, nothing in it asks a court to pull your tracks off streaming services, and I have not seen a single case of a label going after an independent artist for releasing an AI-assisted song. Statutory damages on this scale are a weapon aimed at a venture-backed company, not at somebody with forty tracks on a streaming profile.

That said, I see three real exposures for people like us, and I would rather you hear them from someone who makes music than from a comment section.

The first is platform risk. Udio users learned this the hard way when downloads disappeared as part of its settlement, overnight and without a vote. Suno kept downloads but capped them, and it already retired every earlier model on September 9. Suno says everything you made stays playable and shareable in your library, but the specific model that produced your sound is gone, which matters if you ever wanted stems, alternates, or continuations in the same voice.

The second is that ownership is thinner than the marketing suggests. Under the terms in effect since September 3, Suno assigns its rights in your outputs to you if you are a Pro or Premier subscriber, subject to the commercial use restrictions above. But the same document says Suno makes no promise that any copyright actually exists in an output, and adds that applicable rights holders may have the right to collect revenue when outputs are distributed on outside platforms. Read that last clause twice. It is the sound of the licensed era arriving.

The third is US copyright law itself. The Copyright Office’s January 2025 report on the copyrightability of AI works confirmed that purely AI-generated material cannot be registered, that prompts alone do not make you an author no matter how detailed they are, and that protection attaches only to genuinely human contributions, judged case by case. Your written lyrics, your sung melodies, your arrangement and editing decisions can be protected. The raw generation cannot, which is why the cleanest ownership always comes from recordings humans actually performed.

The Supreme Court closed one door on that front this year. On March 2, 2026, it declined to hear Thaler v. Perlmutter, which left in place the rule that a work generated entirely by AI can’t be registered. As Reed Smith’s analysis of the cert denial explains, works that combine AI material with clear human creative contributions can still qualify, but only the human-authored parts are protected, and you have to disclose which parts AI generated when you register.

The exposure I actually worry about most is smaller than any lawsuit: a single output that sounds too close to a real song. That’s a track-level problem, and it can happen with any generator. In the GEMA ruling, the Munich court found that simple prompts led Suno to produce outputs substantially similar to the original works. If one of your tracks has a hook that makes people say “wait, isn’t that…,” you have a problem regardless of who wins in Boston. And you carry it alone. As Terms.law’s breakdown of Suno’s output rights points out, Suno doesn’t promise its outputs won’t infringe anyone’s rights and doesn’t indemnify users, and its terms make you solely responsible for the outputs you generate.

A quick word for readers outside the US, because the picture abroad is tougher, not easier. The Munich Regional Court ruled against Suno on July 31 in a case brought by GEMA, Germany’s music rights society. Denmark’s Koda sued in November 2025, and Canada’s SOCAN filed a claim on September 2. If your instinct says Europe will be the soft landing for AI music, the courts so far disagree. One point in European users’ favor: analysis from ConductAtlas suggests EU and UK consumers may be protected by unfair contract terms laws that limit how much risk a company can push onto users for outputs they don’t fully control. That protection doesn’t exist in the same form in the US, so American readers shouldn’t count on it.

Which Distributors Will Take Your Suno Tracks

None of the distributor policies I checked mention the September lawsuit, and Universal’s separate suit against DistroKid, covered below, is about undisclosed AI slop rather than honestly credited releases. What matters more is a set of rules that shifted earlier in the year.

DistroKid accepts AI music as long as you hold the rights and avoid impersonation, infringement and spam uploads. Since April 2026, its upload form asks whether AI generated any part of the track and which parts, such as lyrics, music, or all or part of the audio. That answer feeds Spotify’s AI credits, which began rolling out on April 16, 2026, starting with DistroKid. As Songprint’s explainer on Spotify’s AI policy notes, Spotify has said disclosure doesn’t trigger downranking or removal. I’ve ticked that box on every AI-assisted release and seen no difference in performance. Hiding it is what costs you.

TuneCore is the interesting one, because it flipped. Its parent company Believe announced on April 30 that it was blocking distribution of tracks made on unlicensed platforms, and Suno was on that list. Then, in the Believe and Suno partnership announcement on September 8, the company said all tracks created with Suno’s new industry partner model are now eligible for distribution through Believe and TuneCore. The way I read that, v6 tracks are fine at TuneCore and older tracks made before v6 may still fall under the April block. If you’ve got a v5 song you want to release through TuneCore, check with them before you assume it’ll go through.

CD Baby has had the strictest reputation of the big three. A 2026 comparison from Musci described it as blocking fully AI-generated music outright. Read CD Baby’s current guidelines before uploading anything Suno made.

One more trap. As RouteNote points out in its guide to distributing Suno music, a track can be fine for streaming and still be a bad candidate for YouTube Content ID, the system that matches uploads against registered audio and claims revenue. Raw generations can share patterns with other people’s generations, which invites messy disputes. I keep raw generations out of Content ID entirely.

The Streaming Platforms Are Tightening at the Same Time

The lawsuit is only half the squeeze. The places your music actually lives have spent the past year building filters, and you should know exactly where the lines sit today.

Spotify laid out its AI protections on September 25, 2025: an impersonation policy under which vocal imitations of a real artist are allowed only with that artist’s authorization, a spam filter aimed at mass uploads, duplicates, and artificially short tracks, and support for AI disclosure credits built through DDEX, the standards body that defines how music metadata moves between labels, distributors, and streaming services. Music Business Worldwide reported that Spotify had removed more than 75 million spammy tracks in the previous twelve months. Crucially for us, Spotify was explicit that the disclosure system is not designed to punish or bury artists who use AI responsibly and say so. Fill in the fields honestly and you are fine.

Deezer’s numbers show why every platform is nervous. In figures reported by DJ Mag on July 21, Deezer said it now receives about 90,000 fully AI-generated tracks per day, more than half of all daily uploads, up from 44 percent in April. Listeners barely touch that flood, around one to three percent of streams, and Deezer has flagged 85 percent of AI music streams as fraudulent, pulling them from the royalty pool and removing AI tracks tied to fraud. DJ Mag’s report notes TIDAL has moved to exclude fully AI-generated music from royalties, and Bandcamp banned it outright. The new Suno complaint leans on that same Deezer data to argue the dilution of streaming royalties is a present, measurable harm rather than a future one.

Then there is the distributor front, which turned dramatic three days before the Suno filing. On September 15, Universal sued DistroKid in Delaware, accusing the biggest DIY distributor of deceptive practices and copyright infringement, naming 1,000 recordings for now and calling them the tip of the iceberg. Universal was careful to say the case is not about distributing clearly disclosed AI music, and that distinction is your protection, so take it seriously. The complaint leans on a striking statistic from SIQA’s first quarter report: of 1,551 AI tracks submitted to its charts, 90.4 percent were made with Suno and 75.8 percent went through DistroKid.

The industry’s answer is standardization. On September 14, the global recorded music body IFPI launched the Streaming Integrity Initiative, a voluntary set of anti-fraud principles that includes know your customer checks, meaning identity verification for the people uploading music, plus measures to identify AI-generated content. The three majors signed. DistroKid, Believe with TuneCore, and UnitedMasters had not at launch. Wherever you distribute, expect identity checks and AI disclosure fields to become normal, and expect the DistroKid suit to speed that up.

My Honest Read After Two Years of This Fight

I have been rebuilding Suno sketches in my DAW, a digital audio workstation, the software you record and mix in, since the early versions, and I have followed every twist of the Suno lawsuit since the first complaint dropped. So here is my read, clearly labeled as opinion.

This is a negotiation conducted in court filings. Warner, BMG, Believe, and the Udio deals show the endgame the majors actually want: licenses, revenue share, watermarks, and control over downloads, not a dead platform. Suno refused to give up downloads the way Udio did, and this second suit looks like both the price of that refusal and the leverage for the next round of talks. The 60,202 works also serve a colder purpose. If fair use goes against Suno in the first case, the labels now hold a damages number big enough to dictate settlement terms.

The poisoned tree theory is the part I would genuinely watch, because it is untested and reaches far past music. And do not underestimate what Audible Magic changed here. The era of AI companies shrugging that nobody can really know what is in the training data ended the moment fingerprinting ran against Suno’s corpus in discovery.

That discovery playbook will not stay in one Boston courtroom either. Fingerprinting a training corpus is cheap, the majors own the reference catalog, and every AI company that scraped first and planned to apologize later now knows exactly how its data will be audited. Expect the next wave of complaints, in music and beyond, to arrive with match counts instead of educated guesses.

I would not bet on a quick resolution. Suno’s September 1 answer accused the majors of anticompetitive behavior extending an unlawful monopoly over music, which is not the tone of a company preparing to fold, and a $5.4 billion valuation buys a lot of billable hours. But I also would not bet a career on today’s terms surviving contact with the next settlement. Udio’s users woke up one morning to a platform that no longer exported files. That is the real risk profile: not a shutdown, but a stroke of a pen changing what your subscription means.

How I Changed My Own Workflow After September

None of this made me quit. It made me more deliberate.

I start with generation hygiene. I never put a real artist’s name, a song title, or someone else’s lyric into a prompt. I describe mood, tempo, instrumentation and era instead. It feels limiting for about a week, then you stop noticing.

Next comes a similarity pass before anything gets downloaded. I listen against the songs the track reminds me of, and I run a few sections through a song recognition app. If the chorus melody tracks a known song too closely, I either rewrite it myself or throw the generation away. It’s the most useful habit I’ve built.

Then I keep a paper trail. Every possible release gets a folder with a screenshot of my active plan, the generation date, the model version, my prompts, my lyric drafts, and the project file from my DAW. If anyone asks how the track was made, I can show them in five minutes. The model version matters more now, since tracks made before and after September 9 sit in different positions with some distributors.

After that, I add human authorship on purpose. At minimum I write the lyrics, and on anything that matters I replay the key parts and sing or hire a singer. It sounds better and gives me something I can register.

Finally, I disclose honestly at upload. I fill in the AI credits truthfully and specifically. If I wrote the lyrics and sang the vocal over an AI-generated instrumental, I say exactly that.

Questions to Ask Before Your Next Release

The right move depends on what you’re doing with the music. Ask yourself these in order.

Is this for personal listening, a social clip, or a demo? Then the lawsuit barely touches you. Keep making things.

Is this a streaming release under your own artist name? Go ahead on a paid plan, but do the similarity pass, keep your paper trail, and disclose AI use at upload. Pick a distributor whose current policy clearly covers your situation, and remember that TuneCore’s acceptance is tied to v6 tracks.

Is this for a client, a brand, an ad, a film, or a game? Raise the bar a lot. Clients often want a guarantee that the music doesn’t infringe anyone, and Suno doesn’t give you one to pass along. The review from AI Tools Directory makes a similar point, suggesting businesses using Suno output for brand or client work get a current legal view first. For paid sync and commercial work, I either rebuild the track with human performances or I don’t use the generation at all.

Was the song made on the free tier? A free trial download carries no commercial rights, so don’t release that file. Under the September terms, rights come from a download made on a paid plan, so if you love the song, download it again while subscribed before you release it, or rebuild it by hand.

Should you jump to another generator to escape the legal mess? Udio settled with Universal and Warner, but it’s a more closed platform, and Sony filed a fresh lawsuit against it in July 2026 covering more than 30,000 recordings. No major generator is free of litigation right now. Choose the tool for its sound, its workflow and its terms, not because someone online called it safe.

What I Would Do This Week

Enough analysis. Here is the checklist I am personally working through, and I would suggest you do the same while everything still works the way it does today.

Start by getting your library out. Go through your Suno account, decide which tracks actually matter, and spend downloads on them now, prioritizing anything already released, anything promised to a client, and anything you might build on later. Archive the files in two places, and log the date, the model version, and the prompt for each one.

Then document your human contribution on everything you care about. Save lyric drafts, voice memos, arrangement notes, and session files with timestamps. Under the Copyright Office’s approach, those human elements are what you can register and defend, and in any future dispute, contemporaneous evidence beats memory every time.

If a song earns real money or anchors your artist identity, get it out of the AI only basket entirely. Replay the parts yourself, or bring in session musicians and a producer to record the track again as a human performance. A re-recorded master has a clear human author, a full copyright you can register and license, and none of the caveats buried in Suno’s terms. It costs money and time, and for throwaway tracks it is overkill, but for the songs that matter it is the one clean fix available right now.

Disclose your AI use through your distributor’s DDEX credit fields, keep your release volume sane, and stay far away from anything that imitates a real artist’s voice, because authorization is now the explicit rule on Spotify and takedown is the consequence.

Finally, watch how the Boston court handles fair use now that fact discovery in the first case closed on September 30. Those rulings will shape this second Suno lawsuit, the licensed platforms, and the whole AI music economy far more than any press release. The free ride era is over. The era of making good music and owning it properly is just getting started, and that one, at least, rewards the people who treat this like a craft.

Sources

Common questions

Did Suno really get sued again?

Yes. Universal Music Group and Sony Music Entertainment filed a second complaint against Suno on September 18, 2026 in Boston federal court. It covers 60,202 sound recordings the labels say Suno copied without a license, and it exists because a judge would not let them add those works to the original 2024 case.

How much money could Suno owe the labels?

US law allows up to $150,000 per work for willful infringement, which comes to just over $9 billion across the 60,202 recordings. The labels also want up to $2,500 for each act of getting around YouTube's download protections, roughly another $150 million, plus a court order stopping Suno from using their recordings.

What does the poisoned tree claim actually mean?

Suno says v6 was trained from scratch without Universal or Sony recordings. The labels answer that v6 learned from the outputs and user preference data of older models that were trained on their music, so in their view the infringement carries straight through into the new models.

Why is Warner Music not part of the new lawsuit?

Warner settled its claims against Suno in November 2025 and signed a licensing deal. BMG and Believe followed in 2026, and all three are now partners on the v6 models, which is exactly why only Universal and Sony are still suing.

Can I still use Suno while the lawsuit is going on?

Yes, the platform is running and v6 is live. The case is about how Suno built its models, not about individual users. Just be aware that prices, terms, and features have changed several times over the past year and will probably change again.

Do I own the songs I make on Suno?

If you pay for Pro or Premier, Suno assigns you its rights in the songs you download, but its own terms say the company makes no promise that any copyright exists in an output. In the US, purely AI generated music cannot be copyrighted at all, though the parts you genuinely authored, like written lyrics or added performances, can be protected.

Can I still release Suno songs on Spotify?

Yes, if you are on a paid tier and you spend a download on the track first. Free tier songs are for personal use only and cannot go on streaming services. You should also disclose the AI involvement through your distributor's credit fields and avoid imitating any real artist's voice.

What are Suno's download limits now?

Since September 3, 2026, Pro subscribers get 20 song downloads a month and Premier subscribers get 60, with the option to buy more. Free users get seven downloads total, cleared for personal use only. A song cannot be released, sold, or licensed until a download has been spent on it.

Will Suno shut down because of this lawsuit?

Not any time soon. Suno raised $400 million in June 2026 at a $5.4 billion valuation and is fighting the claims in court. The bigger risk for users is not a shutdown but sudden changes to features and terms, which is what happened at Udio after its settlement.

Can I get sued for using Suno songs?

The current lawsuits target Suno, not individual users, and nobody has gone after creators for ordinary releases. However, Suno's terms make you responsible for your outputs and it doesn't indemnify you. The biggest practical risk is a specific track that sounds too close to an existing song.

Does TuneCore accept Suno music now?

TuneCore's parent company Believe said in September 2026 that tracks made with Suno's new industry partner model are eligible for distribution. Songs made on older models may still fall under the block Believe announced in April, so check before uploading.

Should I switch from Suno to Udio because of the lawsuits?

Switching won't take you out of the legal picture. Udio has settled with Universal and Warner but is a more closed platform, and Sony filed a new lawsuit against it in July 2026. Pick a tool for its sound, workflow and terms rather than a promise of safety.