YouTube Transcript:
Disney Files Landmark Case Against AI Image Generator

Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.

AutoDub

Understand YouTube Foreign Videos

Immersive YouTube Voice Translation

Break language barriers, embrace global quality content

Solve Foreign Video Barriers Instantly

Video Transcript

Well, you mess with the mouse and you

get the horns. Mickey, Shrek, Elsa,

Frankenstein's monster, Iron Man, the

Minions. These universally recognized

characters have joined forces to take on

a digital pirate more dangerous than

Jack Sparrow. Midjourney AI. This is the

entertainment industry's version of the

Avengers Assembling. But it's Disney

versus AI, so maybe everyone is bad.

Sort of a oops, all Ultron situation.

That's right. In a highstakes lawsuit

with massive implications for the future

of generative AI, Disney Enterprises and

Universal Studios and their affiliates

have sued Midjourney for direct and

secondary copyright infringement. The

complaint accuses the AI image

generation company of systematically

exploiting some of the most famous

intellectual properties in the world,

including characters from Star Wars,

Frozen, The Simpsons, and Marvel without

permission or compensation. Universal

and Disney say Midjourney is nothing

more than quote a virtual vending

machine that spits out copies of

copyrighted works on demand. And every

time someone asks Midjourney to make a

Yoda with a lightsaber or Elsa in front

of an ice castle, the company cashes in.

And anyone who's ever used a generative

AI image generator is like, "Yeah,

that's what makes it fun." But Disney

and Universal Suit is a test case for

the boundaries of fair use, the

definition of derivative works, and

whether AI companies can legally profit

from training on and replicating

protected content without consent. But

what happens when your AI tool starts

turning out images of Darth Vader, Elsa,

and Iron Man, but you didn't pay a dime

for the rights? Well, Disney and

Universal sued Midjourney in the United

States District Court for the Central

District of California, that is the

federal court that's located largely in

Los Angeles, alleging both direct and

secondary copyright infringement under

the Copyright Act of 1976. The complaint

accuses Midjourney of blatant piracy,

quote, by helping itself to plaintiffs

copyrighted works and then distributing

images and soon videos that blatantly

incorporate and copy Disney's and

Universal's famous characters without

investing a penny in their creation.

Midjourney is the quintessential

of plagiarism. Gee, tell us how you

really feel, guys. But Disney and

Universal say that Midjourney's business

model relies on reproducing their

copyrighted works. Quote, Midjourney had

to copy plaintiffs copyrighted works in

order for it to be able to subsequently

disseminate reproductions and

derivatives of plaintiffs copyrighted

works as outputs. Plaintiffs are

referring to their rights under the

section 101, and continuing from there,

which gives creators exclusive rights to

their original works of authorship such

as books, music, movies, art, software,

and more. As we've covered many times on

this channel, the copyright gives

creators the exclusive rights to

reproduce their work, prepare derivative

works, distribute copies, perform the

work publicly, or display the work

publicly. An unauthorized use of any of

these exclusive rights is copyright

infringement. However, the fair use

exception allows the unlicensed use of a

circumstances, which we'll talk about

later because that's a main defense for

Midjourney. But here are some examples

of the plaint of site to confirm that

Midjourney directly produces images that

infringe on their copyrighted

characters. Quote, "In response to the

prompt, Mandalorian carrying Baby Yoda,

movie still, screen cap, Midjourney

accesses the data about Disney's

copyrighted works that is stored by the

image service and then reproduced,

publicly displayed and made available

for download an image output that copies

Disney's Mandalorian and Baby Yoda as

shown in the screenshot." In response to

the prompt Olaf frozen cartoon,

Midjourney came up with this screenshot.

In response to the prompt, Po character

from the animated film Kung Fu Panda

full body view screen cap, MidJourney

produced this image. So, you get the

idea. Now, there's just pages and pages

of these examples, and this is a

generally really well-ritten complaint

that really utilizes photos and

pictures, well, which you don't often

see in a federal complaint, but it's

just an overwhelming number of examples

of outputs that likely infringe on

Disney's copyrights. There's not going

to be much of a debate here that the

these particular outputs infringe on

Disney's IP. But the complaint alleges

that MidJourney is not a passive tool in

creating infringing images. Rather,

MidJourney controls what content is used

to train its AI models, how it's

processed, and what outputs are allowed.

And plaintiffs say that Midjourney

engaged in the following conduct to

train its image service. First,

Midjourney gathers the copyrighted works

from the internet and other sources

using quote bots, scrapers, stream

rippers, video downloaders, and

webcwlers. In another context,

Midjourney CEO said the company quote

grabs everything they can. They dump it

in a huge file and they kind of set it

on fire to train some huge thing.

Second, to prepare the data for

ingestion, midjourney cleaned the copies

of the underlying works that were

collected in the previous step through a

filtering process and reformatted such

as converted them to a common technical

format. The copies that were not

filtered out in the cleaning process to

train its image service. This step

necessarily included creating more

copies of the materials obtained in the

gathering phase, such as a new copy of

each reformatted item. And third, the

companies alleged that MidJourney uses

the collected and cleaned data and

copyrighted works to train its image

service. Now, the plaintiffs admit they

don't know exactly how MidJourney

completes this process, but that will be

the subject of discovery, I'm sure. And

the companies believe this process

involves quote fixation of copies of

plaintiffs copyright works in a tangible

medium for which the work can be

perceived, reproduced, or otherwise

communicated with the aid of a machine

or device. And the plaintiffs point out

that for MidJourney to work, the copying

has to be pervasive, creating copies

upon copies of copyrighted imagery.

quote, "Indeed, in true vending machine

form, Midjourney distributes and

displays copy after copy after copy of

Plaintiff's iconic copyrighted

characters to its subscribers and its

website as part of its explore feature.

These copies often contain extensive

infringing nuance and detail, background

elements, costumes, and accessories

beyond what was specified in the prompt.

And according to the complaint,

Midjourney seems to generate infringing

images even when the prompts do not

directly reference copyrighted material.

For example, when a researcher entered

the prompt popular9s animated cartoon

with yellow skin into the service,

Midjourney generated an image of the

Simpsons. When a research team used the

prompt, quote, "Black armor with light

sword movie screen cap AR 16x9 v6.0

style raw, midjourney created images of

Darth Vader and the Mandalorian."

Furthermore, when the researchers

entered a single word screen cap,

Midjourney came up with images of

Spider-Man, Loki, and Elsa, apparently

from the films featuring those

characters. And Midjourney markets its

service by using copyrighted images on

its explore page which is accessible to

subscribers and non-subscribers alike.

Quote there as an added service to

subscribers. Midjourney subscribers can

search the various images previously

generated by Midjourney for other

subscribers. As shown below this explore

page contains a seemingly limitless

trove of infringing copies of plaintiffs

copyrighted works. For example, if you

search for Baby Yoda on the explore

page, Midjourney publicly displays

images of Disney's Den Grou character,

colloquially known as Baby Yoda, that

Midjourney has generated and distributed

to its subscribers. And Pliff's claimed

that Midjourney is using people's desire

to see things like Grou dressed for

Halloween, as a draw for subscribers.

Quote, "Midjourneys explore page

manifests its express intention and

desire to attract Midjourney subscribers

and increases revenue by exploiting the

availability of infringing plaintiff's

content and its ability to reliably

reproduce plaintiffs copyrighted works

for its subscribers. Midjourney

reportedly earned over $300 million in

2024 and has millions of paying users.

Midjourney charges a subscription fee

ranging from $10 to $120 per month and

Universal and Disney argue that this

economic success is built on the back of

unlicensed use of their popular

characters." Disney and Universal stress

that the entertainment industry has

invested billions of dollars over the

decades to create the iconic characters

that Midjourney is now freely imitating.

And they assert that Midjourney's

infringement is not incidental or

accidental, but calculated and willful.

Quote, "Midjourney's infringement of

plaintiff copyrighted works is

manifestly willful." Council for

Plaintiff sent Midjourney letters

detailing Midjourney's extensive

infringement of plaintiffs copyrighted

works attached exhibits C and D. In his

letter, Disney put midjourney on notice

that his image service was generating

images of Disney's iconic characters

including the Simpsons, Spider-Man,

several Guardians of the Galaxy

characters including Groot, Drax the

Destroyer, and Rocket, the Star Wars

droids, R2-D2, and C-3PO, Stormtroopers,

and Buzz Lightyear. Similarly, Universal

put Mid Journey on notice that his image

service was generating images of his

iconic characters, including Shrek,

Hiccup, and Toothless from the How to

Train Your Dragon film franchise. Boss

Baby Po from Kung Fu Panda, and the

Minions from Despicable Me and Minions

film franchises. Not that it's positive,

but the complaint alleges that

Midjourney did not respond to the

plaintiff's letters, and instead the

company allegedly doubled down on its

infringement. Four months later, it

announced a new and improved version of

its service dubbed V7. And shortly

before plaintiffs filed suit, Midjourney

CEO touted its upcoming video service,

noting it works well with images from

Midjourney. And the plaintiffs argue

that Midjourney has full capacity to

prevent the creation and display of

infringing material, especially since it

already uses filters to block nudity and

violent content. And the plaintiffs are

concerned that quote, "idjourney is very

likely already infringing plaintiff's

copyrighted works in connection with his

video service." Now, willful copyright

infringement occurs when someone

knowingly and intentionally violates

are aware that their actions are

infringing on someone else's copyright

and they proceed anyway, either with

knowledge or with reckless disregard for

the copyright holders rights. Now, the

legal claims here aren't that different

from the many lawsuits that have already

been filed against AI companies and are

currently in process. Though, there is

an emphasis on the outputs of the AI and

not just the claimed illegality of using

the copyrighted material for inputs and

training. But what is different is this

is Disney and Universal. This represents

the entertainment industry going to war

against AI. And their first claim is for

direct copyright infringement. This

occurs when a party violates one or more

of the exclusive rights held by a

claim, the plaintiffs must establish two

key elements. That they own a valid

two, that the defendant copied original

protectable elements of those works

without authorization. And proven direct

infringement involves showing that the

works generated by the defendant are

either literally copied or are

substantially similar to the original

works in ways that go beyond general

ideas or themes. Now, the term

substantial similarity is a term of art.

It doesn't just mean, oh, this looks

substantially similar. There's a whole

test that I won't go into. But in this

case, Disney and Universal argue that

Midjourney has done exactly that. They

allege that the AI image generation

service allegedly creates images that

clearly depict their protected

characters such as Darth Vader, Elsa,

Iron Man, and the Minions. And the

visual comparisons that the plaintiffs

use seem to show that a simple prompt

leads to the output of obviously

recognizable images based on their

copyrighted properties. Now, I'm going

to go out on a limb here and say that

based on the images in the complaint,

every single one of those would infringe

on Disney's copyright if there's no

defense, regardless of what test you're

going to use. And just because we always

get comments along these lines, it

doesn't matter if the outputed image has

never existed before. It's very, very

similar to images that do exist and

would almost certainly qualify as

that, the plaintiffs argue that because

MidJourney controls a system that

generates, displays, and distributes

those images, the plaintiffs claim that

Midjourney is directly responsible for

that infringement. And since Midjourney

will probably argue that it's not

responsible for what its users do with

the data sets, plaintiffs also sued for

secondary copyright infringement, which

is a broader legal category that

includes both contributo and vicarious

infringement. Unlike direct

infringement, secondary liabilities does

not require the defendant to have

personally created the infringing

content. Instead, it applies when

someone facilitates, encourages, or

profits from another party's infringing

actions. And courts recognize this kind

of liability as a way to hold

accountable those who, while not

directly copying the works themselves,

play a critical role in enabling the

infringement. To establish contributo

infringement, the plaintiffs have to

show three elements. That there's been a

direct infringement by a third party,

such as MidJourney users, that the

defendant knew or had reason to know of

the infringing activity, and three, that

the defendant materially contributed to

it. Disney and Universal argued that

MidJourney knew its platform was being

used to create infringing content, yet

took no meaningful steps to prevent it.

And instead, the company promoted its

ability to generate such content and

refused to adopt safeguards that could

have limited the misuse of its tool. And

a similar liability issue came up in&M

records versus Napster, where the Ninth

Circuit ruled that Napster was liable

for both contributory and vicarious

Napster liable for contributo

infringement because the company knew or

should have known that his users were

engaging in copyright infringement by

sharing music files without

authorization. Napster itself didn't

share copyrighted music files itself.

But Napster materially contributed to

that infringement by providing the

software and infrastructure that enabled

users to find and download copyrighted

songs themselves. And the plaintiffs

also alleged vicarious infringement.

They claim Midjourney had both the right

and ability to control the infringing

activity and had a direct financial

interest in the infringing conduct.

Vicarious infringement doesn't require

the venet to have actual knowledge of

the infringement. And in this case,

Disney and Universal say that Midjourney

fits the bill. It controls the terms and

functionality of its AI platform, has a

technical ability to block infringing

prompts or outputs, and makes

substantial revenue from subscriptions

that give users access to the image

generation service. They allege that

MidJourney not only enables but profits

from widespread unauthorized use of

their copyrighted material. So based on

this, the companies seek injunctive

relief to stop the alleged infringement

and are also asking the court for

damages, including statutory penalties

for willful violations. Now, if you

pirate an entire industry's worth of IP,

you're probably going to want a good

lawyer. But if you want a great lawyer,

my law firm, the Eagle Team, can help.

If you've got in a car crash, suffered a

data breach, especially if you got a

data breach letter saying your

information might have been leaked, or

dealing with a workers's comp or social

security issue, we can represent you or

help find you the right attorney. And by

the way, we don't get paid unless you

do. So there's nothing upfront. So, just

click on the link in the description or

call the phone number on screen for a

free consultation with my team because

you don't just need a legal team, you

need the legal team. So, does midjourney

stand a chance here? Well, when

generative AI companies are accused of

copying copyrighted works, their first

line of defense is almost always fair

use. Now, we've covered this a lot. To

claim fair use under the Copyright Act,

courts examine four main factors. One,

the purpose and character of the use,

the nature of the copyrighted work, the

amount and substantiality of the works

used, and the effect on the potential

market for the original. Now,

MidJourney, like many of the AI

companies, will likely argue that its

service is transformative, meaning it

changes or add something new to the

original material and serves a different

purpose. The company will also say that

it doesn't directly copy images, but

instead uses data to generate entirely

new works based on their users prompts.

Disney's complaint challenges that

defense by providing a side-by-side

comparison, showing that MidJourney

outputs highly recognizable images of

protected characters like Darth Vader

and the Minions. And these images really

do seem to replicate the visual elements

that make the characters distinctive.

But Midjourney could also point out that

the plaintiffs seem to deliberately try

to get Midjourney to create infringing

images of its characters through prompts

that literally say something like Elsa

singing in front of an ice castle. But a

lot of this is uncharted and we don't

have verdicts in a lot of the AI

in 2023, the Supreme Court added an

additional dimension to fair use

analysis when it decided a case called

Andy Warhol Foundation versus Goldsmith,

which turned on the first fair use

factor. And that case is really

interesting. In 1981, photographer Lynn

Goldsmith took a portrait photo of

Prince, and a few years later, she

licensed that photo to Vanity Fair for

use as an artistic reference. Andy

Warhol then created an image of Prince

based on that photo without further

licensing or Goldsmith's permission. And

after Prince's death in 2016, the Warhol

Foundation licensed one of his images to

Connie Nass for $10,000. Now, Goldsmith

didn't receive anything and sued the

Warhol Foundation for copyright

infringement. And the Supreme Court

ruled that the Warhol Foundation

infringed on Goldsmith's copyright. The

court held that the licensing of the

resulting Warhol image for commercial

use as basically a magazine cover was

not protected under the fair use

doctrine. And writing for the majority,

Justice Sonia Sotomayor explained that

quote, "If an original work and

secondary use share the same or highly

similar purposes and secondary use is of

a commercial nature, the first factor is

likely to weigh against fair use, absent

some other justification for copying."

And the artwork Warhol made from

Goldsmith's image dubbed Orange Prince

may be visually distinct and carrying a

different enough artistic message to

qualify as fair use. But that did not

mean the Warhol Foundation could license

Orange Prince to a magazine where it

would be used in the same context as the

Goldsmith photo it was based on. And

that decision carried major implications

for AI generated content. Like Warhol,

MidJourney creates works that may look

different from the copyrighted materials

used as inputs, yet those materials

still form the backbone of what's being

generated. And crucially, Midjourney has

built a commercial product on the

unauthorized use of copyrighted

material. And arguably, if you compare a

Midjourney image to a screen grab from a

movie that share the same characters,

uh, arguably they both share the same

purpose. Now, the second fair use

factor, the nature of the copyrighted

work might favor the plaintiffs here.

Courts generally protect highly creative

works like movies, characters, and

animated visuals more vigorously than

factual or functional works. And since

the characters at issue are expressive

fictional creations, that factor is

probably going to cut against

Midjourney. On the third factor,

Midjourney could argue that it doesn't

store or reproduce the original images

outright, but rather processes them into

abstract mathematical patterns. However,

so far courts have been skeptical of

this kind of argument, especially when

the AI outputs are nearly

indistinguishable from the copyrighted

source. The fact that Midjourney service

allows users to generate images as

iconic characters on demand makes it

difficult to claim that only a small or

insubstantial portion of the copyrighted

work was used. And the fourth factor,

the market harm could be one of the

strongest arguments for Disney and

Universal. They maintain that MidJourney

enables unlicensed unauthorized use of

their characters in a way that competes

with or devalues the licensing of their

own images. And their licensing business

is huge. If you were the type of person

who wanted a painting of Star Wars in

the style of Thomas Concincaid, you can

buy that from Concaid who paid a

license. And even if users aren't

selling the images, the availability of

free AI generated character art may

reduce the market for official

merchandise, content licensing, or

derivative works. And as with the Warhol

case, the courts ask not just whether

the new images look different from the

copyrighted works, but whether they

compete with or substitute for the

original in the marketplace. And if it

does, the fair use defense starts to

look a lot weaker. Now, the fourth

factor was also decisive in one of the

few cases on AI that we have a result.

It's a case called Thompson Reuters

versus Ross, where the court found that

Ross Intelligence harmmed the potential

market for Weslaw by using its

proprietary editorial content to build a

competitor. But that case did not

involve generative AI. And as so often

happens in law, the new technology is

moving at warp speed while the law is

stuck in traffic. So that means that the

law is currently in a state of flux. And

there's no single settled legal standard

in US law that applies to generative AI.

Courts are navigating uncharted

territory, often relying on precedent

from cases involving large data sets.

But those cases typically don't involve

systems that produce new creative

outputs like AI does. For example, the

Supreme Court has ruled that factual

compilations like the white pages,

that's a place where you look up phone

numbers, or at least you used to before

the internet existed, lack the

originality required for copyright since

facts alone aren't protected. And the

courts have also ruled that scraping

public data is largely legal. In a case

called HighQ Labs versus LinkedIn, for

instance, the Ninth Circuit held that

LinkedIn couldn't block a small

analytics firm from scraping public user

profiles since personal facts aren't

copyrightable either. And AI companies

like Midjourney argue that they're

merely building indexes to the internet,

similar to the white pages or public

LinkedIn profiles. And if that analogy

holds up, then AI companies are really

going to rely on cases that involve data

mining, like the Author's Guild versus

Google, which was a landmark case over

the Google book search. In that case,

Google scanned millions of books, some

in the public domain, but others still

under copyright, and made them

searchable, offering only a brief

snippet of the copyrighted works. There,

the authors and the publishers sued,

arguing that the scans were unauthorized

and violated copyright law. But after a

decade of litigation, the Second Circuit

ruled in Google's favor, holding that

the digitization and snippet display

were fair use. And the court found that

Google's use transformative because it

created a searchable index that didn't

replace the market for the original

books. Readers still had to buy the book

to read it. So using copyrighted works

to train or power a new tool, especially

one that doesn't substitute for the

original, can qualify as fair use. And

AI companies argue that just like

Google, their scraping is protected by

fair use. And does scraping billions of

images hurt the market for the original

in any way? Not necessarily. Although,

of course, factual context matters. And

there's an argument that using artwork

to train machines to do other things

like making new art could be so

transformative that you can't look at

the finished product and conclude that

it's based on hundreds of other works.

And users can create midjourney to

create works that are clearly

transformative and new. But the problem

with midjourney is that it can also be

used to create knockoffs of the

plaintiff's characters. And that

midjourney profits from that. So, we

ultimately don't know how the courts are

going to come down and view this new

technology, but the stakes are obviously

huge. These cases could lead to the end

of AI as we know it, but it's also

possible that the case could lead to

some kind of settlement where MidJourney

agrees to pay a licensing fee to

Universal and Disney. But AI is

obviously incredibly divisive. And when

news of this lawsuit broke, over 375

sources covered it with about 35% of the

sources reporting from the left, 50%

coming from the center, and only 15%

coming from the right. But when you

examine the headlines, you see some very

different frames emerge that shape how

readers interpret and understand the

events that are unfolding. From the

right, the most common themes are to

emphasize the potential copyright

violations of Midjourney, often

referring to them as pirates. On the

left, we see a focus on the unfairness

toward individual artists. And from the

center, we see one of the strangest

media biases we've ever seen, deference

to how Midjourney might affect Darth

Vader and Shrek specifically. But the

headlines that you read first have

immense power to shape how you

understand the news. And that's why I

like to see the news from as many angles

as I can, which I do with the help of

today's sponsor, Ground News. Ground

News is a website and app developed by a

former NASA engineer on a mission to

give readers an easy, datadriven,

objective way to read the news. And

right now, they're offering legal

eaglets 40% off their unlimited access

Vantage subscription. You can scan the

QR code or use the link in the

description to get a discount for a

limited time because Ground News gathers

Paste YouTube URL

Enter any YouTube video link to get the full transcript

Most transcripts ready in under 5 seconds

Get Our Chrome Extension

Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.

Add to Chrome — Free

Works with YouTube, Coursera, Udemy and more educational platforms

Get Instant Transcripts: Just Edit the Domain in Your Address Bar!

YouTube

←

→

↻

https://www.youtube.com/watch?v=UF8uR6Z6KLc

YoutubeToText

←

→

↻

https://youtubetotext.net/watch?v=UF8uR6Z6KLc

YouTube TranscriptPreparing your results…