Should AI companies pay photographers? Yes. If a model trains on your photos and sells what it makes from them, that is commercial use of your work. Commercial use gets paid. The only honest model is ongoing royalties. When your images help generate revenue, you get a cut of that revenue. Anything less is theft with better branding.
I have been a working photographer in Los Angeles for 27 years. I have shot celebrities, campaigns, and editorials. I know exactly what it costs to build a style: years of bad light, missed moments, and frames that taught you something. So when I say it takes years to develop a style and to be ripped off by a computer in an instant is insulting, I am not being dramatic. I am describing the economics of my profession.
This is not a theoretical debate for me. It is my livelihood. And the courts are finally starting to agree that something has to give.
## What is the Midjourney copyright lawsuit about?
In June 2025, Disney, NBCUniversal, and DreamWorks sued Midjourney. Their complaint called the image generator a "bottomless pit of plagiarism." The studios argued that Midjourney reproduces recognizable characters and copyrighted work without permission and without any real safeguards.
Eighteen months later, nothing has changed. A report published October 14, 2026 by VN.ai, a startup that stress tests AI tools for copyright infringement, named Midjourney the worst offender for flouting Hollywood copyright. The company that was sued for mass infringement allegedly kept doing it.
Read that again. Sued by three of the biggest studios on earth. Called a bottomless pit of plagiarism in a federal complaint. And a year and a half later, an independent audit says it is still the worst in the business. That tells you everything about the incentive structure. Theft is free. Lawsuits take years. The math favors the thief.
## Is training AI on photos copyright infringement?
That is the question every court in the country is wrestling with, and photographers should pay attention to the answers coming in.
The biggest case from working artists is Andersen v. Stability AI, filed by illustrators who discovered their work had been scraped into training data. That case has survived multiple rounds of motions to dismiss. A summary judgment hearing is scheduled for February 2027, with a trial set for April 2027. The artists are still standing. The platforms have not made this go away.
Then there is Getty Images v. Stability AI, which matters for a different reason. In November 2025, the UK High Court ruled that Stability AI's use of Getty's trademarks inside generated images was trademark infringement. More importantly, the court held that the model provider, not the user, bears responsibility. Read that carefully. The company that built the model is on the hook, not the person who typed the prompt.
That ruling cuts straight through the industry's favorite defense. AI companies love to say they are just a tool, like a camera. A camera does not ingest five billion images and sell the statistical residue. When the provider is responsible, the provider has to pay. That is the direction the law is moving, and photographers should be cheering it on.
## What would fair pay actually look like?
Most people in this debate stop at "pay the artists" and never say how. That vagueness is convenient for the platforms, because a demand without a mechanism is easy to ignore.
Here is my mechanism: ongoing royalties. If an AI company uses my images to create other images, and those images generate revenue, I get a cut of that revenue for as long as it flows. Not a one time buyout. Not a flat licensing fee that undervalues the work. Royalties.
This is not a radical idea. Musicians get royalties when their songs play. Photographers already license images with usage based pricing. [What a celebrity photographer actually costs](https://www.dclee.net/blog/how-much-does-a-celebrity-photographer-cost) depends on usage, because usage is where the value lives. AI training is usage. Massive, industrial scale usage. Price it that way.
The platforms will say this is technically impossible, that you cannot trace which training images influenced which outputs. That is their engineering problem, not mine. They built systems that ingest the entire visual internet. Figuring out how to pay for it is the cost of doing business. If your business model only works when the raw material is free, your business model is broken.
## Someone stole my photos once. I sued them and won.
This is not theory for me. I have lived the small version of this fight, and I want working photographers to hear how it went.
Years ago I shot an editorial of a luxury condo building in downtown Los Angeles. It was a real assignment. I scouted it, lit it, and delivered the work. Six months later I saw my pictures being used in an ad for the building. Nobody asked me. Nobody paid me. They just took the images and ran a campaign with them.
I had a lawyer send a cease and desist. They ignored it and kept running the ad. So I filed in small claims court.
And I won. They paid me.
I tell this story for one reason. The system works when you use it. Most photographers never file. They complain to friends, post about it, and let it go because lawyers are expensive and court sounds intimidating. Small claims court exists precisely for this. The filing fee is small. You do not need a legal team. You need your contract, your images, and the evidence that they used them.
Now scale that up. If one photographer can beat one condo developer in small claims, imagine what happens when thousands of photographers stop accepting theft as the cost of doing business. The AI companies are counting on our silence. Every artist who fights makes their math worse.
## Why should anyone outside photography care?
Because photographers are just first in line. The same fight is coming for every creative profession.
An analysis by economist Joseph Politano found that the United States has lost more than 200,000 jobs across creative industries in the past four years. Film and TV alone lost more than 100,000 jobs, nearly a third of the sector's workforce. Written publishing lost over 70,000. Politano is careful about how much of that is directly caused by AI versus consolidation and offshoring, and he is right to be careful. But the timing lines up with the generative AI boom, and nobody serious thinks these forces are unrelated.
Look at what is happening inside photography specifically. The AI headshot market exceeded $350 million in 2025 and is projected to pass $450 million in 2026, according to Narkis.ai. That is nearly half a billion dollars of revenue built on a simple proposition: why hire a photographer when software can fake the result? If you want the honest comparison between the two, read my breakdown of AI headshots vs a real photographer. The short version is that software can copy a look but cannot direct a human being.
Here is what the platforms do not want to discuss. Every AI headshot generator was trained on real photographers' work. The poses, the lighting patterns, the retouching styles. All of it came from people who spent careers learning those things. The $450 million market exists because photographers spent decades building the visual language it copies. Royalties are not charity. They are back pay.
## What can photographers actually do?
First, register your copyrights. In the United States, registration with the Copyright Office is a prerequisite for bringing an infringement lawsuit. Some of the artists suing the AI platforms had claims dismissed simply because they had not registered. Do not give the other side a free procedural win. Register the work.
Second, document everything. Keep your RAW files, your contracts, and your delivery records. My condo case was winnable because I could prove the images were mine and prove they were used without permission. Paper trails win cases.
Third, stop giving your work away for exposure. Every free image you post without terms is training data with your name scrubbed off. License your work like it has value, because it does.
Fourth, support the cases that are already moving. Andersen v. Stability AI goes to a summary judgment hearing in February 2027. The Getty cases are grinding forward on two continents. These rulings will set the terms for the next decade. Pay attention to them.
And fifth, say it out loud. I am a celebrity photographer with 27 years behind the camera, and I am telling you plainly: AI companies should pay photographers for copyright infringement. The more working professionals who say it on the record, the harder it gets for the platforms to pretend this is a fringe complaint.
## The bottom line
Midjourney got sued by Disney and kept going. An independent report says it is still the worst offender. The courts are slowly catching up, from the UK trademark ruling to the Andersen trial in 2027. The direction is clear, but direction is not payment. Payment comes when photographers demand it, document it, and sue for it when they have to.
It took me years to develop my style. A computer ripped it off in an instant and the industry called it innovation. I sued a condo developer in small claims court and won. The AI companies are bigger, but the principle is identical. My work has value. Use it, pay for it.
Should AI companies pay every photographer whose work trained their models, with ongoing royalties for as long as that work generates revenue?