AI music lawsuits: How businesses can stay safe in 2026
AI music lawsuits are no longer a rarity in 2026. Learn how to stay safe.
The creative industry is moving fast. Almost too fast. Generative AI has changed how music is created, edited, and used in video content, encouraging speed and experimentation. For brands, broadcasters, and publishers, AI can be a powerful tool.
But it also comes with real legal complexity.
As AI music tools become part of everyday life, regulators, courts, and platforms have started closing in on the legal gray areas.
In 2026, the message is simple: creativity and speed are still possible, but only if the legal foundations are solid.
We’ll discuss:
- AI music legal issues: The basics
- Legal developments shaping the AI music landscape in 2026
- Breaking down 4 key AI music lawsuits
- What’s next for AI music?
- What this means for brands, broadcasters, and publishers
- Use Epidemic Sound to find, edit, and clear music safely
AI music legal issues: The basics
AI doesn’t change the fundamentals of music rights. If anything, it exposes how flimsy commonly held assumptions can be.
When the origin or licensing behind a piece of music is unclear, the risks pile up. Content takedowns, blocked campaigns, platform strikes, unexpected royalty claims, lost reach — it’s all on the table. Such risks impact creative teams trying to move fast just as much as legal teams tasked with protecting the brand.
Some of the most common risk areas include:
- Unclear training data and infringement risks: Some AI music models were trained on datasets without licenses in place. If an output is found to include parts of, or closely resemble, an existing track, liability doesn’t stop with the AI provider. It also extends to the brand using the music in content.
- Gaps in ownership and control: Not all AI-generated music can be owned or protected. In many markets, fully AI-generated tracks are not eligible for copyright protection. That makes it difficult — or impossible — to secure exclusivity or prevent others from using the same music.
- Responsibility doesn’t disappear: Even when music is generated by a tool, brands remain responsible for how it’s used. This means they still need to ensure the music can be synced to video, distributed commercially, and published across platforms under the license.
Legal developments shaping the AI music landscape in 2026
Over the last few years, a series of high-profile protests, lawsuits, and settlements have begun to shape the legal landscape around AI music. However, much remains unresolved, and the legal framework is still taking form.
Who’s suing?
Lawsuits have been initiated by record labels, publishers, performing rights organizations (PROs), and individual creators. The rights holders claim that the AI companies have trained their models on copyrighted recordings, compositions, and lyrics without a license or any remuneration. Some have characterized it as “the biggest theft in music history.”
Who’s being sued?
AI music generation platforms Suno and Udio have been hit by several lawsuits, but more “general” AI companies like OpenAI and Anthropic also face copyright infringement litigation. The claims are twofold: unauthorized use of copyrighted training data, and in some cases, that the AI-generated outputs reproduce protected works.
Breaking down 4 key AI music lawsuits
Let’s break down four key AI music lawsuits, looking at how they have shaped — or will shape — the legal debate and landscape moving forward:
- GEMA sues OpenAI (ChatGPT) for reproducing copyrighted lyrics
- Suno and Udio lawsuits in the US
- Suno lawsuits in the EU
- Anthropic’s training data comes under scrutiny…again
1. GEMA sues OpenAI (ChatGPT) for reproducing copyrighted lyrics
GEMA, the German performing rights organization and one of the largest in the world, brought a case against OpenAI. It claimed that ChatGPT’s models had been trained on protected song lyrics without a license, and that ChatGPT could reproduce those lyrics almost word-for-word when prompted.
The court ruled in GEMA’s favor in November 2025, finding that OpenAI had copied, stored — not simply studied — and made available copyrighted works. OpenAI has announced its intention to appeal, and the judgment is not yet final.
The ruling made clear that the AI provider — not just the end user prompting and using the output— has a responsibility when models reproduce protected content.
2. Suno and Udio sued in the US
In 2024, the Recording Industry Association of America (RIAA) filed lawsuits against AI music platforms Suno and Udio. Representing the major labels Universal Music Group, Sony Music Entertainment, and Warner Music Group, the RIAA argued that Suno and Udio’s models were trained on copyrighted recordings without permission, and that the outputs those models produced could replicate and reproduce protected recordings closely enough to infringe copyright.
By late 2025, Suno settled with one and Udio with two of the three major labels. Suno signed a licensing deal with Warner Music Group, while Udio struck deals with both Warner Music Group and Universal Music Group. Under the agreed terms, both platforms committed to move toward licensed training data, launch new models, and enforce tighter usage controls. Warner Music Group’s agreements included download caps on Suno, plus a full ban on downloads and external sharing on Udio.
In May 2026, Sony and Universal Music Group asked the court to add more sound recordings to their lawsuit against Suno. Their initial 2024 complaint listed 560 works, which rose to “millions” after using audio fingerprinting technology to identify further infringements. Of those millions, Sony and Universal Music Group are pursuing 61,026. In the same month, Sony requested that 30,442 more copyrighted works be added to its case against Udio.
3. Suno lawsuits in the EU
European collecting societies have also taken legal action against Suno. Germany’s GEMA filed against Suno at the Munich Regional Court in January 2025, alleging its models were trained on works from its catalog without a license, and that outputs replicate identifiable copyrighted compositions.
Denmark’s KODA filed a separate lawsuit against Suno in Copenhagen in November 2025. The lawsuit alleged unlicensed training on KODA’s members’ works, citing recordings by artists including Aqua, MØ, and Laid Back. Both the GEMA and KODA cases are ongoing.
4. Anthropic’s training data comes under scrutiny…again
Anthropic had already dealt with a lawsuit from book authors, for which it paid $1.5 billion to settle: the largest copyright settlement in US history.
In 2023, a coalition of music publishers including Universal Music Publishing Group, Concord Music Group, and ABKCO filed a copyright infringement lawsuit against Anthropic, alleging that song lyrics had been used without a license to train Anthropic’s chatbot, Claude. That initial filing covered around 500 tracks and sought approximately $75 million in damages.
In January 2026, the publishers expanded the case significantly, adding over 20,000 tracks and raising the potential statutory damages sought to more than $3 billion.
No settlement has been reached at the time of writing. In March 2026, a group of US music-industry bodies including the RIAA, NMPA, the Music Artists Coalition, and SoundExchange filed an amicus brief to support the publishers.
What’s next for AI music?
While cases make their — often slow — way through the court system, regulation, licensing, and platform policies are all moving in parallel. Let’s look at how the law, AI companies, and rights holders are building stronger foundations for the future.
More transparency requirements
Rights holders and regulators are pushing for clearer disclosure around how AI content is created and licensed. One example is the EU AI Act, which forces developers to publish summaries of training data, and requires that AI-generated content is marked as such.
Moreover, different types of attribution technology are being developed to identify, trace, and quantify the copyrighted material used to generate AI music. These tools could help artists and rights holders understand when and how their work has been used in AI-generated music, and support attribution and remuneration. PROs are also actively assessing how usage of AI music can be tracked, attributed, and monetized.
Labeling obligations being introduced
Various labeling requirements are being introduced, both under law and under platform terms and conditions. Under Article 50 of the EU AI Act, anyone deploying AI to create so-called “deep fakes” is required to clearly disclose that the content is AI-generated or manipulated.
Platforms like TikTok, YouTube, and Meta have all introduced mandatory labeling requirements for AI-generated content, backed by automated detection systems. The obligations vary from platform to platform, so creators and businesses need to check each of the different policies to stay compliant.
Clearer rules around voice and identity
The legal protection of voice and likeness has become increasingly important. Both Taylor Swift and Matthew McConaughey have filed trademark applications to protect their voice and image against AI misuse.
In the US, the proposed federal NO FAKES Act is in the legislative process, but has not yet been signed into law. The NO FAKES Act would establish a federal right of publicity in the US. This would protect artists’ likenesses from being used without their consent, give them control over their digital personas, and hold companies and individuals liable for producing, distributing, or posting unauthorized digital replicas.
Certain artists have also licensed their voices on their own terms, like world-famous actor Sir Michael Caine, who has licensed his voice to ElevenLabs.
Partnerships between AI companies and rights holders
AI is here to stay, and some of the trendsetters have already made deals with rights holders. Suno and Udio’s newly licensed ecosystems suggests that the era of “unlicensed scraping” could be wrapping up and replaced by label-approved guardrails.
However, these platforms are far from fully licensed. They still need to find resolutions with rights holders, like publishers and PROs.
What this means for brands, broadcasters, and publishers
For brands, broadcasters, and publishers, tighter AI music laws mean fewer gray areas and less tolerance for unclear licensing. Here’s what it could look like.
Businesses need the right to use the music, not just generate it
Creating a track with AI doesn’t automatically grant the right to use it in content. Always confirm whether the terms allow commercial use, distribution, and synchronization to video; and who is responsible if a copyright claim arises. Platforms and rights holders will notice and act upon infringement, especially when a commercial aspect is in play.
Exclusivity and control still matter
If you can’t own or protect the music your business uses, you can’t control how it appears alongside your brand. That lack of control can derail flagship campaigns and long-term assets, undermining the brand’s credibility along the way.
Risk will remain
As technology and the legal landscape evolve with rapid pace and scale, it’s hard to keep up. Until fair, robust licensing and remuneration agreements are in place between gen-AI music platforms and rights holders across the board, businesses may still expose themselves to risk for unlicensed usage of output.
Social media usage may still prove tricky
Social media is a hotbed for music lawsuits — businesses are frequently found in breach of copyright, whether accidentally or intentionally. Digital fingerprinting and detection will only increase in the wake of AI music.
Businesses will look for safe, straightforward, and ethical music licensing
AI-generated music introduces unresolved legal and ethical questions. The legal landscape may become clearer at some point, but right now, businesses need reassurance. They need to know all the necessary rights are included for their use cases, and that they’re cleared forever.
Don’t risk your business with a risky license.
Use Epidemic Sound to find, edit, and clear music safely
Epidemic Sound owns all the rights to the music in its catalog, each of the 55,000+ tracks composed and performed by real artists. That means a clean chain of title, no hidden rights holders, and no unlicensed or opaque training data.
And if you want to experiment? Use AI-powered tools like Adapt, Assistant, and Studio in tandem with original tracks. Find, edit, and integrate Epidemic Sound’s music into your projects with AI. Online and off, for your brand’s and clients’ channels. Safely.
Regardless of platform and media switch-ups in the future, your content is secure. What you publish during an active subscription is licensed forever, even if you choose to leave Epidemic Sound.
The Enterprise plan is the perfect solution for modern businesses and workflows. Epidemic Sound’s dedicated digital rights management team works tirelessly to ensure licensed users don’t encounter issues when publishing content. They’ll walk you through the licensing process, providing you and your team with the tools you need to use music safely.
Everyone from Levi’s to Warner Bros. Discovery, The Telegraph to the Miami Dolphins, trusts Epidemic Sound to help tell their stories. Enjoy global coverage and tell your brand’s story, your way.
If you run a small to midsize business, try the Epidemic Sound Business plan instead.