AI and Copyright: What the Recent Lawsuits Actually Mean for Artists

If you've been following the news, you've probably seen headlines declaring that artificial intelligence companies have "won" the copyright battle.

That's not what happened.

In fact, the recent wave of lawsuits involving AI, books, artwork and newspapers show something very different: courts are beginning to draw distinctions between different uses of copyrighted works rather than issuing a blanket approval or rejection of AI training.

The legal landscape is becoming more nuanced, not less.

That’s important because AI and copyright is complex and deserves to be peeled back and analyzed closely to ensure that new precedents are fair and strong.

As photographers and artists, it's worth understanding what these cases actually say because the headlines often leave out the most important parts and your work may be at issue now, or in the future.

The AI Copyright Cases Are Not All Asking the Same Question

One of the biggest misconceptions is that there is a single "AI copyright case."

There isn't.

Instead, courts are being asked to answer very different questions depending on the type of work involved.

Some lawsuits focus on books.

Others involve photographs and artwork.

Some involve newspapers.

Others concern music or video.

Each raises different legal issues, even though they're often grouped together under the label "AI lawsuits."

Images: Artists vs. AI Image Generators

The lawsuits involving image generators such as Stability AI and Midjourney are among the most significant for photographers and visual artists.

These cases generally ask questions like:

• Was copyrighted artwork copied during AI training?

• Were copyrighted images stored on company servers?

• Can AI-generated images infringe existing copyrights?

• Should AI companies obtain licenses before training on artists' work?

• 
  

One of the most closely watched developments came when Disney and Universal sued Midjourney, alleging the company used copyrighted characters and imagery without permission and generated unauthorized reproductions of iconic characters. The studios are seeking damages and an injunction, arguing that AI does not receive a free pass simply because software is involved.

These image cases remain ongoing and will likely shape how visual artists' works are treated for years to come.

But it is important to consider: if these companies don’t just view what’s available online, if they store the imagery on their servers to train their models, does that increase their participation and liability—was it willful and does that infringe?

Is it different than someone storing an image on their computer to use as a reference for a painting but creating something different? How different? Are they copies or derivatives?

If an AI company doesn't simply access publicly available images online but instead copies and stores millions of copyrighted works on its own servers for the purpose of training a commercial model, does that additional act of copying carry independent legal significance?

Does creating a permanent training dataset constitute infringement in itself? If the copying was intentional, what role, if any, does willfulness play?

The June 2025 decision in Bartz v. Anthropic illustrates why these questions matter. While the court held that Anthropic's use of lawfully acquired books to train its language model could qualify as fair use under the specific facts before it, Judge William Alsup drew an important distinction between the training process and the acquisition of the underlying works. He concluded that Anthropic's retention of millions of pirated books downloaded from unauthorized sources was not protected by fair use, allowing those copyright claims to proceed. The decision suggests that courts may analyze how copyrighted works are acquired and stored separately from how they are ultimately used to train an AI model.

Another question artists frequently ask is whether AI training is akin to a human artist studying reference images.

An artist may legally look at thousands of paintings, photographs, or illustrations throughout a lifetime, absorb techniques, composition, color, and style, and later create an original work influenced by those experiences. Copyright law has long recognized that ideas, techniques, and artistic inspiration are not protected. Expression is.

But is that what an AI model is doing?

Or is it creating and retaining digital copies of millions of copyrighted works before mathematically encoding them into a commercial system? If so, is that more similar to a person downloading and maintaining a massive archive of copyrighted images than simply viewing artwork for inspiration?

The answers remain unsettled today. It might be answered in the future in one ruling  and then it might be contradicted in the next.

That’s how I see this going because AI is so complex it will take years to set solid foundations of precedent that is consistent enough to guess how some future cases may end, but in the legal world, nothing is guaranteed as we’ve seen in the last couple years with landmark cases being overturned and new precent setting cases being approved at the Supreme Court level.

Future courts will likely continue examining where the line falls between learning from copyrighted works, making copies of them, creating derivative works, and generating outputs that may themselves infringe existing copyrights. Those distinctions may ultimately shape not only the future of generative AI but also the scope of copyright protection in the digital age.

I don’t think that scope is close to being carved out as one might hope. The digital age we live in is fast-paced and the new technologies are hard to keep up with in courts that need time to marinate on pleadings and while they do that,  new things will pop up and so on it will go.

At least some players in this AI landscape will have job security....haha hmmm.

Books: The June 2025 Decisions That Sparked the Headlines

Much of the recent media attention came from two federal decisions involving authors and AI companies.

The first one I mentioned above.

In Bartz v. Anthropic, Judge William Alsup concluded that using copyrighted books to train Anthropic's language model was a transformative fair use under the facts before him. However, the court also held that Anthropic's creation of a massive library containing pirated copies of books was not fair use and allowed that portion of the case to proceed.

Around the same time, Kadrey v. Meta resulted in a summary judgment for Meta, but not because the court declared AI training universally lawful. Instead, the court found the plaintiffs had not presented sufficient evidence of market harm on the record before it.

Those are important distinctions.

And consider that perhaps its too soon to gather the evidence of market harm by AI. While it consumes the headlines, generative AI is still relatively new. We need a deep understanding of impact and harm so that information can be shared in cases, as plaintiffs or defendants.

So the long and short:

Neither decision announced that all AI training is legal.

Neither decision held that creators have no rights.

Both decisions were heavily dependent on the evidence presented in those particular cases.

Newspapers: A Different Kind of Copyright Battle

The lawsuits brought by newspapers, including The New York Times against OpenAI and Microsoft, raise different concerns than the book cases.

Publishers argue that AI systems do more than learn from journalism. They contend that AI generated answers can substitute for visiting the original news source, reducing website traffic, subscriptions, advertising revenue, and licensing opportunities. Because news derives much of its commercial value in the hours and days immediately following publication, publishers argue that AI summaries and responses may capture that value without bearing the cost of producing the reporting.

The lawsuits also allege that AI models have, in some instances, reproduced portions of copyrighted articles rather than merely learning from them. If proven, that presents a different legal question than simply using published works for training.

Also, you might wonder why you see the same article language used across multiple newspapers and if they all infringe upon each other anyway, but they actually license the use of the same article language, which is income for the original publication.

That fact highlights an economic reality. The publishers argue that AI companies are benefiting from the same reporting without obtaining comparable licenses. Whether copyright law ultimately requires such licensing for AI training remains one of the central questions the courts must answer.

These cases will likely help define how copyright law addresses not only AI training, but also market substitution, licensing, and the reproduction of protected expression in AI-generated outputs.

Unlike a novel that someone may read once, newspapers rely on ongoing commercial value from timely reporting. That makes questions of market substitution especially important.

The courts have not yet resolved these issues, making the newspaper cases some of the most significant copyright disputes still pending.

Video: The Next Frontier

I think video may become the next major battleground.

As AI systems become capable of generating increasingly realistic video content, copyright owners in the entertainment industry are beginning to ask many of the same questions photographers have been asking for years:

• Can copyrighted films be used for AI training?

• When does generated content become infringing?

• What responsibilities do AI companies have to prevent obvious copying?

Many of these questions have yet to receive definitive answers, but the lawsuits now being filed suggest the next several years will be pivotal.

What These Cases Mean for Photographers

For photographers, one takeaway is especially important:

The recent decisions did not eliminate copyright protection for photographs, nor did they hold that every use of copyrighted work for AI training is automatically fair use.

Instead, courts are carefully examining:

• how copyrighted works were obtained,

• how they were used,

• whether they were copied or stored,

• whether the resulting AI outputs compete with the originals,

• and whether creators suffered measurable market harm.

Those are highly fact-specific questions that impact a final decision.

At the same time, a growing issue has emerged within the photography community. It's increasingly common to see posts calling out photographers or businesses who allegedly upload another creator's image into an AI platform with instructions such as, "remove the people," "change the background," "make it look like this," or "recreate this scene." The resulting image is then shared or marketed as if it were an original creation.

These practices are often defended with statements like, "AI made it," "I changed enough," or "it's just inspiration." However, those claims are not legal defenses in themselves. Whether an AI-assisted image infringes copyright depends on the specific facts, including how much protected expression from the original work was used, how closely the new image resembles the original, and whether it qualifies as a lawful derivative work, fair use, or an independent original creation.

As courts continue to address AI-related copyright disputes, photographers should not assume using AI automatically avoids infringement simply because software, rather than traditional editing tools, was used. Copyright law focuses on the resulting use of protected expression, not merely the technology used to create it.

The Bottom Line

The story isn't that AI "won."

The story is that courts are beginning the difficult work of applying long-standing copyright principles to rapidly evolving technology.

Some AI companies have won important rulings.

Creators have also won important rulings.

And many of the most significant questions, particularly those involving photographs, artwork, newspapers, and video, remain unanswered.

If anything, the recent decisions show that copyright law is not standing still. It is evolving one carefully reasoned case at a time.

For artists, photographers, writers, and other creators, this is not the end of the conversation.

It's only the beginning.

AS ALWAYS:

Disclaimer

I am not an attorney, intellectual property professional, or legal advisor. This article reflects my understanding of recent copyright cases based on publicly available court filings, judicial opinions, legal commentary, and my own research.

More importantly, it reflects my perspective as a professional photographer, artist, and small business owner whose copyrighted work has been infringed multiple times. Like many creators, I've spent years navigating copyright enforcement, DMCA takedowns, licensing disputes, and legal proceedings firsthand.

Nothing in this article should be considered legal advice or a substitute for consulting a qualified attorney. My goal is simply to share information, encourage thoughtful discussion, and help fellow creators better understand the rapidly evolving intersection of artificial intelligence and copyright law.

Image Credit: The header image used in this article is a fully licensed stock image made available through the Wix Media library. As a photographer, artist, and copyright advocate, it's important to me to be transparent that I did not create or photograph this image. It is used in accordance with Wix's licensing terms. I would have preferred to use one of my own photographs, but due to storage limitations under my current website plan that was not possible.