Suno Lost to GEMA. Is AI Copyright Entering a New Era?
Disclaimer: As usual...I am not an attorney or intellectual property professional. This article reflects my personal opinions and observations as a photographer, artist, and small business owner whose copyrighted work has been infringed multiple times. It is not legal advice.
Just a few weeks ago, much of the conversation surrounding AI copyright centered on Bartz v. Anthropic.
Many headlines boiled the decision down to one sentence:
"AI training is fair use."
As I wrote in my previous blog, I don't think that's the takeaway from Bartz.
Now another case has entered the conversation, and it points in a very different direction.
A German court recently ruled against AI music company Suno in a lawsuit brought by GEMA, the organization that manages rights for composers and music publishers in Germany. According to reports, the court concluded that Suno's use of copyrighted music for AI training required authorization under German law. Suno has announced its intention to appeal.
So...
Does this mean AI companies lost? Meh, not exactly. But I like the way the court looked at it.
That's the way it goes in the legal world. The court, the judge, the district, the details...they all bring about different results when put together, which means I don't think there will ever be "ONE" copyright case we can rely on to guide us in our own legal endeavors, it just gives us little nods to specific details that may or may not impact our own cases.
There May Not Be One AI Copyright Rule
For the past two years, everyone has been looking for one answer to one question:
Is AI training legal?
And I think that's the wrong question.
Instead, we need to ask a bunch of questions before we get our answer:
What type of copyrighted work is involved?
How was it obtained?
Was it licensed?
Was it copied?
Is there an existing licensing market?
Does the AI output compete with the original work?
Which country's copyright law applies?
Why Music Is Different
Music has something many creative industries already understand well: licensing.
For decades, businesses have licensed music for radio, television, streaming, restaurants, films, advertising, and live performances. Organizations like GEMA exist specifically to manage those rights and ensure creators are compensated.
That makes the Suno case particularly interesting.
Rather than asking whether AI can "learn" from music, the lawsuit asks whether companies should first obtain permission when using copyrighted songs to build commercial AI systems.
That feels like a very different conversation than whether someone can simply listen to music and become inspired by it.
Compare That to Bartz
In Bartz, the court analyzed books.
More specifically, it separated how the books were acquired from how they were ultimately used for training.
The court found that training on lawfully acquired books could qualify as fair use under the facts before it, while also making clear that pirated books were an entirely different matter.
The Suno decision, while applying a different country's laws, reminds us that courts are not all approaching AI training the same way.
Personally, I prefer the way the Munich court looked at it. As I said in my Bartz blog, if content is being used specifically for commercial purposes, even if that is training AI models, then the creators of those works should be compensated through use licensing.
Afterall, the industry of arts has a nasty little phrase we just can't get away from: Starving Artists. Sadly, we often starve each other.
Not compensating artists for their art should never be an option.
What Does This Mean for Photographers?
As photographers, it's tempting to look at every new AI case and ask:
"What does this mean for my images?"
The honest answer is...
I still don't know.
Photographs are not books.
They're not songs.
They're licensed differently.
They're consumed differently.
And image-generation models raise their own unique questions about copying, memorization, derivative works, and market substitution.
The lawsuits involving Getty Images, Stability AI, Midjourney, Disney, Universal, and groups of visual artists are still working through those issues.
I don't think we should assume the outcome of a music case automatically predicts the outcome of an image case. But it sure is getting interesting.
My Take
The future of AI copyright isn't going to be decided by one case, or one judge and probably not even one country.
Instead, courts appear to be building the law one factual scenario at a time.
Books.
Music.
Newspapers.
Photographs.
Film.
Each creative industry has different licensing practices, different markets, and different ways in which AI can affect creators.
Personally, I don't see that as a weakness.
I see it as courts doing exactly what they're supposed to do: examining the facts before making sweeping legal rules.
As someone whose own photographs have been infringed, I still believe creators deserve meaningful protection in the age of generative AI.
Whether that protection ultimately comes through courts, legislation, licensing systems, or a combination of all three remains to be seen.
But if Bartz taught us that how copyrighted works are acquired matters...
The Suno decision reminds us that what is being copied, where it's being copied, and how creators are traditionally compensated may matter just as much.
And I like the way they think, I hope more courts start considering us more.
