"I was searching this pirated database of books that have been taken off the internet, and it was bought by Meta in 2020..."

That's how our story begins. Through Ann Handley's newsletter, I discovered an Atlantic article with a way to search the pirated LibGen database containing everything from Harry Potter to Stephen King — even Anne's books. A database Meta purchased in 2020.

Search LibGen via The Atlantic

And it reveals something critical about how AI companies approached training data.

A Meta director of engineering admitted in a 2020 chat (revealed through legal filings):

"The problem is that people don't realize that if we license one single book, we won't be able to lean into fair use strategy.

So we'll have to drop all of the LibGen and books data sets."

2020 - Meta Director of Engineering

This engineering choice wasn't just about cost—it was about time. Those pirated books fueled early AI development before most people even knew what was happening.

It's cheaper to risk lawsuits than to start paying everyone. That's the reality we're facing.

But here's the contradiction that nobody's talking about:

Today, the creative economy is worth an estimated $985 billion. G20 Insights estimates that by 2030, the creative economy could account for 10% of global GDP.

Deloitte predicts that by 2030, there will be up to 40% growth in the creative industry.

If AI is "stealing" all creative work, how is the creative economy projected to grow so dramatically?

The creative industry is rising. So should you.

This contradiction sits at the heart of our conversation today. The creative industry isn't dying—it's transforming.

And instead of fighting yesterday's battle, let's explore how creators can position themselves over the next 2-5 years when everything's changing.

"The AI freight train is coming on.

I'm saying to the creators, here's what you can actually do."

Part 1: Creators as Energy Feeding AI

"It starts with knowing the side hustles."

Instead of fighting for micro-payouts after the fact, what if creatives became direct contributors—fueling AI with expertise, unique insights, and high-quality input?

This isn't about being replaced by AI, but about becoming an essential part of AI's lifecycle.

Most of the AI innovation and experimentation is happening at the big enterprise level—big tech, large companies where major experiments are underway.

I've been dissecting these down to see what might apply to you and me.

"What if we could fuel AI with expert insights, knowledge, experience and also creativity, that dynamic part of humanity, not just limit the probability and the creation to AI?"

Small Side Hustles: Specialized Data Curation

Financial services, human resources, marketing and sales—these three industries have been the first broad lines of AI adoption because they involve more predictable scenarios like creating words or formulas.

Still AI models lack niche, high-quality domain knowledge—especially in:

* Highly specialized writing and creative workWhen a friend asked me about selling copywriting services, I told him, "I don't think people are looking for copywriting services. They're looking for your knowledge, your experience, your words using AI." That's exactly what he does—creating a voice, a tool, their own AI copywriter built with them.

* Cultural context"I live in Northern California, which most people consider Los Angeles. But if you know the United States, we're part of Cascadia—Washington, Oregon, and the upper northern part of California. People barely ever visit." AI often lacks this deep social understanding and cultural point of view. It only knows what we've written about.

* Authentic voices and visual knowledge"People who are photographers, people who are artists can be able to use their visual knowledge, their domain-specific knowledge to help people work with GenAI." Knowing all the aspects that a photographer understands to get that good shot—the kind of film, the camera, the lighting—these are specialized skills AI still needs help with.

The best thing about these side hustles? The AI training takes time and usually is more than a month or two.

Plus, "if there's no training and it just ends, there's entropy. It erodes. Your service might be something you could offer again and again."

Think of it as teaching AI to speak the language of your creative domain—combining technical understanding with creative intuition to help businesses find their way through this new landscape.

Part 2: AI-Driven Creative Careers

"This is future looking. I don't like to do that.

But if we're seeing a creative economy growing so much,

where might some of the opportunities be?"

Beyond side hustles, entire new career paths are emerging with creativity and AI.

These aren't just theoretical futures—they're starting to take shape right now.

AI Creative Residencies

"Let's go back to the Renaissance."

In the Renaissance, artists weren't paid to sell their books—it was hard. The Medicis, for example, that famous family (sort of like The Sopranos of their time), worked with Michelangelo as a patron, creating art.

"What if we had these micro-Michelangelos, this ability to distribute this knowledge and creative residencies where this creative person and their domain would come in for three months, six, 12 months, maybe hired in-house from bigger companies to help their training?"

AI companies (or even third-party organizations) could hire creators to:

* Fine-tune AI models with unique styles and develop those styles with them

* Create AI-enhanced experiences (what the customer sees when visiting a site)

* Act as AI "consultants" for businesses trying to implement generative AI

"A creative residency allows you to use your creativity, and you can start applying this to small nonprofits, small businesses.

There's all these people who are so lost on AI and so few are helping them."

Instead of viewing AI as defeating creativity, think of it as an opportunity to bring your specific skills to help others understand their customers, their data, and improve it through a creative residency.

"AI is driven by probability. We need a good and a bad to find the middle."

This understanding opens up creative possibilities that technical teams might miss.

AI Agent Definers and Designers

"The marketers have taken over AI agents last year... but here's the thing about AI agents—they're good.

They're really good at doing automations.

They're sometimes good at doing workflows."

A new hybrid role is emerging that blends technical understanding with creative instinct:

* Designing AI personalities and behaviors that resonate with specific audiences

* Creating workflows that combine AI capabilities with human creativity

* Defining how AI interfaces with creative processes

"This isn't programming. It's creative direction for AI behavior." But you need to know a little bit about how this works—take some classes, get yourself familiar, do some prompting.

"Where do people start losing it?

Where is the organization of projects happening and going wrong?

Have you ever been on one where it just goes off the rails because people aren't that organized?"

These are the workflow gaps where creatives bring value.

The most valuable skill isn't just making content anymore—it's defining how AI interprets and generates content within specific creative contexts.

Help people find their way through this new landscape with your domain expertise combined with AI understanding.

Part 3: The Reality Check on Licensing

"Everyone who's a creator says, 'I want to get paid for what was taken from me.' And while that's good, where were you getting paid before AI took it from you?"

While we're all frantically trying to prove AI stole our content, the creative economy has always been a fame lottery.

Only a small percentage of creators make substantial money from their work. AI isn't changing that ratio—it's just changing how the game is played.

The Fame Lottery

"If your content was taken and it made a lot of money, you're going to be one of the many talking about licensing deals or represented by lawyers who are trying to figure this out."

Consider these realities:

* Getty Images watermark shows up in Stability AI outputs—their brand became a pattern because it was repeated so often

* Famous authors like J.K. Rowling might get licensing deals because their work creates recognizable patterns

* For the average creator? "None of us obviously exist at this level"

The reality is, finding your influence in AI models is like finding a specific drop of water in the ocean.

The Licensing Economics

I went and asked a number of different bots and people like, how would this compare to how musicians get paid at Spotify right now?

Right now, AI companies have paid undisclosed amounts to outlets like Axios and Axel Springer for licensing deals, but let’s construct some hypothetical numbers.

If you're a musician on Spotify, there is so little money from all that streaming that you have to go on the road and play concerts because that's where you make your money. Writers and artists who don't have a live stage face an even tougher challenge.

Plus, it's not a good comparison:

"When I go to Spotify, I'm playing your song, or at least a piece of it, right?

But when I'm going to AI, I'm probably maybe getting a piece of your influence."

This leads us to what's called "nonliteral influence," which sounds hokey, but it's essentially about probability putting together patterns. AI isn't storing your exact content—it's reconstructing patterns it's learned.

"For Getty Images, that's brand... If you have your little photos going out there, at best it could partially influence what's going on."

The Licensing Reality

Here's the reality nobody wants to talk about: even if we could perfectly track every creative work that influenced an AI model, the resulting payment would look more like a Spotify royalty than a Hollywood contract.

AI licensing—it's the same. You're a creator. Are you going to be a hit? Are you going to be a star? Are you going to be on top of things? Then you might have some room to talk about licensing.

You get big, you're going to have attorneys and publishers. Those are the ones getting paid.

Part 4: The New Creative Reality

I'm not really talking about fighting as much as adapting.

We still need human creativity and judgment.

The difference is we need it in new places—not just making content, but shaping how AI understands and generates content.

From Content Creators to Data Strategists

The most valuable creative skill isn't just creating content anymore—it's understanding how data shapes AI's perception of quality, style, and meaning.

Creators can:

* Design content that maintains unique human value

* Develop expertise in guiding AI tools rather than just using them

* Focus on what AI can't do well: emotional resonance, cultural relevance, moral judgment

AI Needs Human Creativity

Instead of fighting AI influence, creators can intentionally design content that stands out in an AI-saturated world.

The future belongs to creators who can:

* Use AI to handle scale while focusing on uniqueness

* Leverage AI tools to enhance rather than replace their creativity

* Position themselves as essential interpreters between AI and human audiences

The question isn't whether AI will use your work—it already has. The question is whether you'll find ways to make your creativity valuable in an AI-powered world.

After all, the most valuable creative skill has never been just making content—it's been making content that connects. And that human connection is exactly what AI still can't replicate.

The choice is yours: stand on the sidelines hoping for licensing checks or become part of the system that shapes AI's creative future.

"Breathe and put out your voice. Because we need IT."

RESOURCES

Search LibGen, the Pirated-Books Database That Meta Used to Train AI

Anne Handley: Everybody Writes

The Unbelievable Scale of AI’s Pirated-Books Problem

How The Growing Creative Economy Will Soon Devour The Real Economy

Creative economy 2030: Inclusive and resilient creative economy for sustainable development and recovery

The creative economy takes center stage - 2021

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

"'I've always tried to understand the world underneath everything, because that's really what's going on,

We live on the surface.

And unfortunately, the culture that we live in is also absolutely obsessed with surface... But that isn't reality.'

What you're looking at is more powerful than any legal argument.

On the left, AI-generated "in the style of Geoffrey Laurence." On the right, an actual Geoffrey Laurence painting.

"On first glance you could say, oh yeah, that's a Geoffrey Laurence, but I would never do that in the background.

I would never make a background like that.

Look at the reflected light in the cheek and just below the ear.

Do you see the difference between that and what the machine has done on the left side of his face? How crude that is."

Geoffrey Laurence

This visual difference shows what's at stake in these copyright cases:

* It's not just the final image - it's the expression, the emotional resonance that Geoffrey describes as "the world underneath everything"

The "It's Different This Time" Argument

The Authors Guild lawsuit makes a fascinating distinction:

"ChatGPT and other LLMs are not search engines. They do not lead readers to authors' books; they replace them with automated derivatives that do not compensate the human creators."

This cuts to the heart of what makes the NYT case different too - it's not just about learning from content, but potentially replacing the need for it.

And now we begin with 3 Case Files, and First…

a Case without a file:

Case with no number, Geoffrey Laurence. Artist, 56 years. All his work sucked up by ChatGPT because he shared it on that flat surface called the Internet.

His work shown in museums worldwide yet to the AI bots, he’s just another pixel towards training that diffusion model.

Because AI is “learning”, all good to take whatever content they can scrape buy or borrow.

And it’s the opposite of what I've discovered investigating the three major AI copyright lawsuits.

The headlines focus on copied text, reproduced images, and legal technicalities.

But that's just the surface. We’ve covered those in detail, picking apart the real from the not so real.

Yeah they matter, but to a judge with no precedents and little real guidance, it comes down to one factor they look at, when they look beneath the surface….

Then Swen Werner with a comment suggests I read his Substack My Digital Truth® - Fusionistic

Asking me to look beneath the surface and shows me a Guy Fieri example, the one that went viral, and how it shows the dancing with the facts that happens in legal matters and with attorneys….

First glance you get angry, then you get curious. Then you take action. Sitting still is not in the cards for creators.

When you follow the money trail like I have, these lawsuits tell a completely different story – one that has little to do with Geoffrey or the millions of other creators without powerful lawyers.

Today we're opening three case files that reveal what AI copyright lawsuits are REALLY about.

When we pull apart the legal word salad and find where the actual decisions will happen, you'll usually find it when you follow the money.

Beneath the surface reality of AI is a money trail telling a different story – one defining the future relationship between human creativity and artificial intelligence.

Case #1: Thomson Reuters vs ROSS Intelligence

Exhibit A: The Competitor's Dilemma

Let's open our first case file.

Thomson Reuters, owner of Westlaw, created a valuable legal research system with human-written headnotes summarizing key points of law.

This represents decades of expert legal services – their intellectual property.

ROSS Intelligence wanted to build a competing AI legal research tool. When Thomson Reuters refused to license their content (seeing ROSS as a competitor), ROSS didn't give up.

Instead, they hired LegalEase, a third-party research firm, to create 'Bulk Memos' containing legal questions and answers.

Essentially, ROSS had found a workaround to use Thomson Reuters' data after being explicitly denied permission.

ROSS bought more than the usual legal office would touch in a decade worth of legal memos all at once – 25,000 of them – creating an anomaly so large it couldn't be ignored.

Thomson Reuters didn't need to find their content in ROSS's output.

The court focused on the training data itself – the Bulk Memos used to teach the AI – finding that they resembled Thomson Reuters' original headnotes.

Follow the money trail:

This wasn't about protecting content – it was about protecting market position.

Thomson Reuters faced a direct competitor attempting to leverage their decades of investment without compensation.

That's what drove the decision, not abstract concepts about AI copying.

Why is this sending a chill through Silicon Valley?

Because copyright violations after AI had legally purchased data was found. On the input, which is the key to the AI industry which states it has a right to use this under fair use.

Fair use defense took a big hit in this one.

Case #2: NY Times vs OpenAI

Exhibit B: The Media Model Under Threat

Our second case file reveals something fascinating about the New York Times lawsuit against OpenAI.

The Times claims ChatGPT is essentially stealing their journalism. Their smoking gun? When prompted, ChatGPT reproduced parts of a viral restaurant review about Guy Fieri.

But dig beneath the surface, and you'll find a different story.

As Swen Werner points out, there's a crucial detail the Times left out: ChatGPT only reproduced about 20% of the article – mostly the beginning that went viral across the internet.

What they didn't say was that it only recalled the most widely quoted parts that likely appeared on countless websites.

Is this evidence of direct copying or something else entirely?

The AI isn't 'storing' articles but recognizing text patterns that appear frequently online – similar to how you might recognize 'To be or not to be' without having memorized the entire play of Hamlet.

Follow the money trail:

This case isn't about copying snippets.

It's about the New York Times seeing AI potentially cannibalizing their subscription model by answering questions using information drawn from their journalism.

Their business model depends on being the exclusive source for their reporting – and AI threatens that exclusivity.

The New York Times lawsuit focuses on surface copying while the concern runs deeper: will people still pay for subscriptions in an AI-powered world?

Case #3: Getty Images vs Stability AI

Exhibit C: The Brand Value Heist

Our third case file might be the most visually compelling – and it crosses international borders in ways that dramatically impact the outcome.

Getty Images, a major stock photo provider, is in legal action against Stability AI, claiming they 'scraped' millions of Getty's copyrighted images without permission to train their Stable Diffusion AI model.

Unlike the other cases, Getty has visual evidence that's impossible to ignore – AI-generated images that accidentally reproduce Getty's distinctive watermarks.

It's like catching a forger who accidentally left the original artist's signature on their work.

This case is being tried in the UK, not the US. Stability AI claims the training didn't happen in the UK at all, but in the US, potentially putting it outside UK copyright jurisdiction.

The court acknowledged that a crucial question is the location of the training.

Follow the money trail:

While 11,383 images aren't that many in the grand scheme of AI training data, finding Getty's brand appearing throughout Stability's outputs reveals what's really at stake.

This isn't just about images – it's about Getty's brand value, worth billions, being used without compensation.

Stability argues that their AI doesn't 'memorize' images but learns patterns, creating new works rather than copying.

They even claim any similarity to Getty images comes from user prompts, not the underlying model.

When AI reproduces a Getty watermark, it reveals that the machine doesn't understand what it's copying – it's just pattern-matching without comprehension.

The economic impact is clear: Getty Images watches as AI tools generate images that could eliminate the need for stock photo licenses, potentially undermining their entire business model.

Buying AI Data with Copyright Diligence

These three lawsuits are reshaping how AI companies acquire and use data. The wild west era of "scrape everything" is ending as courts increasingly recognize that:

* Training = Using: Using content to train AI, even without displaying it, can be infringement

* Intent Matters: Building direct competitors with someone else's content faces greater scrutiny

* Economic Impact Drives Decisions: Cases with clear business model impact have stronger claims

Now what do we do with the damage caused, and how does this work going forward?

We're watching three lawsuits that will determine whether AI thrives or stumbles in America - and I'm struck by how both sides are missing what's really at stake.

It's not about algorithms or legal precedents. It's about who gets to profit from human creativity in the age of AI.

* When the New York Times sued OpenAI, they claimed ChatGPT was essentially stealing their journalism.

* The Thomson Reuters case showed that even without visible output, training AI on copyrighted material may still be infringement when building a direct competitor.

A special kind of sneaky move by ROSS - after being denied a license, they bought 25,000 "Bulk Memos" through a third party to train their legal AI.

An agile AI startup meets a big company that watches it’s 3rd party data purchases.

* The Getty Images case provides the most visible evidence - AI art generators that accidentally reproduced Getty watermarks.

But while others might have felt their work was taken by AI, Getty’s brand is what emerges, which makes that one even harder to figure out.

With their recent merger with Shutterstock in January 2025, the value of that company is about $3.5B US.

So all those Getty Images name and logo appearing, whether with an image or not, are more than copyright violations. They impact brand value by putting Getty’s name next to sometimes odd output.

Because like NY Times snippets, the reconstruction of that repetitive brand copyright image became a recognizable pattern to AI.

That’s more than content, that’s AI taking a brand by accident….which is why this case will be so hard to decide. The numbers here are huge.

Add onto the fact that Stability’s location makes copyright a question of which country, US or UK, but the Getty Images brand is worldwide.

When we look beneath the surface of these three case files, a clear picture emerges.

These lawsuits aren't ultimately about copying, or technical details of how AI works – they're about who profits from human creativity in the age of AI.

* Thomson Reuters doesn't want a competitor using their legal analysis to build a competing product

* The New York Times doesn't want AI undermining their subscription model

* Getty Images doesn't want its brand value appropriated without compensation

These economic realities, not abstract legal principles, will likely determine the outcomes.

The Human Cost Both Sides Are Missing

AI engineers often see creators as entitled complainers who don't understand technology, clinging to outdated ways.

Creators see engineers as people who don't respect the thousands of hours it takes to develop creative skills. Or create.

Neither side is truly listening. And they’re not even dealing with the facts, just opinions on whether AI is copying or evil, and whether creators expect to get paid for things they don’t deserve to be paid for….

If engineers win outright, what motivation will people have to create when their work just becomes training material for Big Tech's trillion-dollar AI systems?

If creators win completely, a thriving industry gets derailed, potentially setting America back in the AI race.

The solutions will likely involve:

* Fair compensation systems that recognize when AI benefits from creative work

* Clear attribution when AI builds on specific sources

* Opt-out mechanisms for creators who don't want to participate BEFORE content goes into AI

* Licensing frameworks that make compliance straightforward

Whether you're a creator like Geoffrey Laurence worried about your livelihood or a business excited about AI's potential, these cases will impact your future.

The companies that will lead us forward aren't just those with the best technology, but those building financially sound data practices into their foundations.

These legal battles aren't the end of the story - they're the messy beginning.

"What I find myself in now is a world that is purely engaged in surface," Geoffrey Laurence told me. "I've always tried to understand the world underneath everything, because that's really what's going on."

This is the perfect metaphor for our AI moment.

The companies creating AI are building tools that excel at surface pattern recognition.

But Geoffrey's work reminds us what lies beneath – the human emotion, the 56 years of craft, the ability to capture something ineffable.

The difference between surface copying and deep understanding matters - not just in art, but in how we build our AI future.

That's the money trail worth following.

RESOURCES

Geoffrey Laurence

The NYT’s AI Lawsuit Hinges on a Misleading Claim—And Nobody Noticed

Every AI Copyright Lawsuit in the US, Visualized

What is Third-Party Audience Data? Examples, Datasets and Providers

Getty Images, Shutterstock gear up for AI challenge with $3.7 billion merger

Authors Guild v. Google, Inc., No. 13-4829 (2d Cir. 2015)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

That buzz when you discover something that feels right - that's how I felt about the UK's 'Make it Fair' campaign for creators' rights with AI.

Web site © NMA Ltd’ or ‘Courtesy of NMA Ltd’.

Then boom - nothing. Like creators trying to make up for lost time.

And what’s “AI fair” varies depending on the country you are in…

Like chasing a static object dancing in random neural networks.

Or eating over-ripe sourdough bread in a seedy San Francisco bus stop.

Sourdough bread and AI. What do they have in common?

Both begin with something small that keeps growing, learning, and evolving.

Some sourdough starters are over 100 years old, taking in new flour and water each time, growing into something new. AI does the same with creative work.

But while everybody loves the smell of fresh bread, we're less comfortable with AI consuming our creativity without clear boundaries.

At least Japan gives you rules you can understand. Pretty much every other country is sitting on the sidelines, waiting for courts to figure it out.

The battle for creative work isn't tech versus artists.

It's about whether we value both the starter and what grows from it.

Like sourdough requires regular feeding, AI needs continuous training with fresh data—not just now, but 100 years from now.

In the future that won't be the Wild West AI of the past few years.

That's why the UK's "Make It Fair" campaign launched this week, another voice shouting the same message.

Meanwhile in the courts and regulatory agencies, many decisions will be made this year.

Creative industries in many countries are fighting back against rules letting AI use copyrighted work without permission, payment, or involvement from content owners.

They're not anti-tech, they're pro-creator, demanding the obvious: consent and compensation.

Bakers often share their sourdough starters as gifts.

AI companies in many countries took creative work without asking. We're not saying stop.

Act like a baker, not a unicorn.

We're saying think and ask before you act.

This battle between creators and AI isn't just playing out in studios and corporate boardrooms between lawyers.

It's happening at the international level, with two countries taking completely different approaches to the same problem.

"The battle for creative work isn't tech versus artists.

It's about whether we value both the starter and what grows from it."

And whether what is “Fair” in the UK, is the same in the US, or Japan?

Japan vs. US: Two Models For AI's Creative Future

In Japan, all copyrightable data is allowed to be put into AI. That might sound like a free-for-all, but they've established rules, boundaries, and practices that make this work better than you might expect.

Japan's approach to AI is community-focused, viewing creativity as an evolutionary timeline.

What somebody creates now builds on what came before, like software development. They're encouraging learning and training while still providing protection for creators.

The Japanese model doesn't mean creators have no recourse. Even without lawyers, they offer mediation and arbitration so anyone can get their case heard.

It's not perfect, but it creates a framework where AI development and protection can coexist. And it’s also indicative of the culture of copyright in Japan.

This contrasts with the US approach, where we're still figuring things out case by case, waiting for the courts to establish precedents.

Our uncertainty creates anxiety for both creators and engineers. Both feel a negative outcome would be destructive to their careers and businesses.

Whatever happens, half will feel let down, and possibly left behind.

Japan's copyright rules for AI rest on two key pillars that balance progress with protection:

The Non-Enjoyment Purpose Requirement

Intention matters. Japan allows copyrighted content to train AI systems if the purpose isn't to recreate the original work. You're not bringing in Picasso’s paintings to create more Picassos - you're learning artistic styles, influences, and color theory.

This creates breathing room for AI development while maintaining boundaries around blatant copying. The focus isn't on inputs but outcomes.

Article 34 Proviso: Where Protection Lives

Japan doesn't leave creators without protection. Their Article 34 Proviso acts as a safety net, establishing that if AI output causes actual harm to a creator's career, reputation, or financial status, normal copyright protections kick in.

This practical approach says: let data flow into AI systems, but watch carefully what comes out. If the output creates direct competition or damages creators, that's where the line gets drawn.

“Make it Fair” for who?

Meanwhile, the UK is having its own creator uprising. The "Make It Fair" campaign launched in February 2025 stands in stark contrast to their 2019 rules regarding AI and copyrightable data.

While the 2019 rules allowed AI to use copyrighted data without permission or payment, "Make It Fair" aims to protect creators' rights and ensure fair compensation.

The campaign addresses the broader impact on creative industries, which generate over £120 billion annually for the UK economy.

They're advocating for stronger copyright protection, opposing government proposals that would weaken existing laws.

Unlike the 2019 rules, the campaign emphasizes the need for AI companies to obtain permission and provide payment for using creative content.

"Japan's approach creates breathing room for AI development while maintaining boundaries around blatant copying. The focus isn't on inputs but outcomes."

Fair for creators versus Fair Use for AI is a defining point in legal decisions.

AI as Hero AND Villain: The Joseph Campbell Perspective

I’d love to ask mythologist Joseph Campbell about AI, he'd likely see our collective anxieties as a projection of our shadow selves.

Just as vampires emerged from three people creating horror stories in 1816 and evolved into a cultural phenomenon, AI represents our deepest fears and highest hopes.

Campbell would recognize AI playing dual roles in our modern mythology: hero and threshold guardian.

For AI enthusiasts and engineers, it's the hero's journey personified - overcoming odds, transforming challenges, and bringing about a whole new world.

It represents progress, potential, and possibility.

For creators whose work has been taken without permission, AI represents the opposite - a destroyer, not a builder of worlds.

It's the threshold guardian, that archetypal figure standing at the door between the old world and the new, challenging whether we're ready to proceed.

This tension isn't new. When electronic synthesizers entered music in the 1970s, many musicians feared the end of "real music."

When digital photography emerged in the 1990s, professional photographers predicted the death of their craft (though Kodak was crushed).

When CGI revolutionized filmmaking in the early 2000s, actors worried they'd be replaced by digital avatars (and that’s looking more possible with AI).

Each time, the technology became a tool that expands rather than replacing the human element. The initial fear gave way to integration - but not without establishing boundaries and rules.

Oh yeah, and don’t steal things first without asking.

Silicon Valley finds that a dated practice, and in this case it’s hard to stop the AI train.

That's why creators are so concerned with legacy. If they can't make a living from their work like Dr. Polidori (who wrote one of the first vampire stories), why create at all?

The fear isn't about AI creating - it's about AI regurgitating what humans have already done without adding value or respecting the source.

Our relationship with AI is complex. We’re both drawn to its potential and frightened by its impact. Like all threshold guardians in mythology, it forces us to confront hard questions about what we value and how we define creativity itself.

"AI plays dual roles in our modern mythology - both the hero on a journey to transform the world and the threshold guardian challenging whether we're ready to proceed."

The 3 C's: Creator Control Toolkit

The Creator Control Toolkit is a way to grab back ownership of your creative work in the AI age.

While many creators feel powerless against AI training on their content, there are practical steps you can take right now.

Control - Take Back Your Creative Power

Up until recently, creators have had virtually no control over how their work is used online. People can take, sample, and repurpose your creative materials with little recourse. But that's changing.

* Time/Date Stamp Everything - Whatever you create, get an official timestamp. Services like CredTent can provide certificates or blockchain records that can be updated.

* Set Up Robots.txt - This simple file on your website can block search engines and some AI crawlers from scraping your content. While not foolproof (many large language models ignore it), it establishes your intent.

* Update Your Copyright Notice - Ensure your website explicitly addresses AI training in your privacy policy and terms of service. (I've posted a template on my Substack that you can use - I got it from ChatGPT, it's that easy!)

ChatGPT No AI Training language example:

Not a legal contract, run this by an attorney. I am not one.

No AI Training Clause

* Prohibition on AI Training UseYou may not use, reproduce, or distribute any content from this website, including but not limited to text, images, videos, or other materials, for the purpose of training, fine-tuning, or enhancing any artificial intelligence (AI) models or machine learning algorithms without prior written consent from [Your Company Name].

* Licensing RequestsWe recognize that there may be legitimate uses of our content for AI development. If you are interested in licensing our content for such purposes, you must contact us directly at [Contact Email or Form] to discuss terms and obtain explicit permission.

* Breach and ConsequencesUnauthorized use of our content for AI training or related purposes is a violation of these Terms of Service and may result in legal action, including but not limited to injunctive relief and claims for damages.

* Monitoring and EnforcementWe reserve the right to monitor usage of our content and take necessary action to protect our intellectual property. This includes contacting organizations or platforms that may misuse our content.

* Reservation of RightsAll rights not expressly granted are reserved. This includes the right to refuse any licensing requests at our sole discretion.

If you have questions about this policy or wish to request a license, please reach out to [Contact Information].

* Specify Your AI Training Policy - Clearly state whether you allow AI training on your work. If you do allow it, under what conditions? With licensing? For free? Make this clear.

* Run anything legal by an attorney if you can. I’m not an attorney and this isn’t legal advice. These are simply actions to take to protect yourself.

Consent - Set Clear Boundaries

Consent means having a say in how your work is used. Without establishing boundaries, you're giving permission for your content to be used however AI companies see fit.

* Be Specific About Permissions - Don't leave it to interpretation. State clearly what is and isn't allowed with your content.

* Watch For Services That Assume Consent - Some platforms' terms of service include language that grants them rights to use your content for AI training. Read the fine print!

* Join Collective Licensing - There's strength in numbers. Publishers like Penguin are bringing authors together for collective AI licensing arrangements. Smaller networks are popping up for each creative niche.

* Document Everything - Keep records of your policies and when they were implemented. Screen captures and a Google Drive helps, and similar ones.

Compensation - Get Paid For Your Value

If your work is valuable enough to train AI on, it's valuable enough to be compensated for.

* Explore Licensing Options - Smart contracts on blockchain platforms can automate licensing for your work.

* Consider Collective Bargaining with Providers who may scale - If there are thousands or millions of creators together, you have more leverage.

* Evaluate Compensation Models - Not all payment structures are equal. Consider whether the compensation offered is worth permanently allowing AI to learn from your work.

* Remember: You Can't Turn Back Time - Once an AI has trained on your work, you can't undo it. Be intentional about your decisions.

It's easy to find reasons things won't work, but finding reasons they will begins with protection and understanding the assets you have as a creator.

They say AI output is just a roulette wheel of randomness. That's only true if you let it be random.

Maybe AI gives us an ability to control the roulette - but only if we first control our relationship with AI.

Establish your presence and protection now, before regulations start adding up. As AI gets better, you'll be protected early.

"The three C's for creators aren't just nice ideas - they're your lifeline in the AI era: Control, Consent, and Compensation."

Owning your creative future: making AI Sourdough with engineers

Ever taste 100-year-old AI? How about sourdough bread from Boudin Bakery in San Francisco with a starter from the Gold Rush Era?

Both sourdough and AI rose during a gold rush, are dynamic evolving entities, and need to be refreshed regularly - one with water and flour, the other with data.

AI has a lot in common with sourdough bread.

Sourdough has a starter that can be 100 years old, each time adding more as it keeps growing, learning, and evolving.

AI is similar, but while we all like sourdough bread (at least the smell of it), we don't like what AI's doing.

Part of the problem is we can't see, taste, feel, or touch AI - and we're told that our content fed into it is gone. The way it happened was such a slap in the face to creators.

What I like about Japan's model is they stated their approach clearly. I'm no lawyer and I'm not giving legal advice, but they took an interesting and intelligent approach that protects both AI and creators.

That's what I want to work on with you this year. I think both sides are heading toward the same goal and should work together.

This isn't about stopping AI - it's about making clear boundaries when AI is taking things, when it's replicating, and when it's intended to replace people.

The one thing Japan recognized that Silicon Valley missed is that AI isn't just about technology - it's about community. If you hurt individuals, you hurt the community.

What goes in is what comes out. Garbage in, garbage out. A relationship with your community is what will come out in your AI.

How can we find that copyright middle ground where creators and engineers work together?

Creators bring perspectives you don't see - they're the edge cases personified.

Engineers bring structure, coding, logic - everything. Together?

Smart software companies always rely on their users as a major source of growth, and AI won't be different.

They need creators. The challenge is acknowledging this opens the floodgates to lawsuits and compensation for everyone who's created content online.

It’s the copyright version of mutually assured destruction, zero sum games.

If only one side wins, both sides lose.

What happens when countries take completely different approaches?

Japan is moving forward, allowing AI access to copyrightable content but with clear guidelines that make things easier to navigate than the US approach.

Currently, the US is still figuring out whether AI-generated content is copyrightable and what happens when copyrighted content goes into AI.

This is the core business practice bumbled by scraping, buying, and avoiding copyright until the lawyers come….which only happens when the money comes.

As we grow, each country thinks their AI approach is the right one, but Japan's model recognizes the community – via its AI - needs to learn while protecting individual rights.

The US is making major copyright decisions around this time next year.

Do you think taking copyrighted content without permission should be allowed?

How do we handle permission at the scale of billions of documents?

Japan's model does something smart - it stands up for both AI's ability to use copyrighted materials to learn AND for the people who own the IP and content.

They're doing this case by case, and I think that's where the US is heading, just moving slower.

I don't want us to become exactly like Japan, but I'm keeping an eye on their approach. A lot of what engineers say makes sense - these models can get smarter and do things for us.

Not eliminating people, but handling tasks that aren't worthy of our time, freeing us to focus on what really matters.

Allowing people to do the work, not the busy work.

Lawsuits over IP scraped, bought, and ingested?Busy work trying to avoid the inevitable and costly retrofit until the next round.

"This isn't about stopping AI - it's about making clear boundaries when AI is taking things, when it's replicating, and when it's intended to replace people."

RESOURCES

Make it fair: The government is siding with big tech over British creativity.

Report on AI and Copyright Issues by Japanese Government

NO&T IP Law Update

Artists release silent album in protest against AI using their work

Chinese Court Issues First Decision on AI Copyright Infringement

China’s First Case on AIGC Output Infringement--Ultraman

AI Copyright Tools

Pixsy - Fight Image Theft (founded 2014)

Created by Humans: Jan 2025

AI Rights Licensing Platform for books

Bernstein.io

Web3 solution to secure and leverage IP assets

MUSIC

SongSecure

https://songsecure.com/

CONTENT AND ART

Credtent.org

=============================

AI Optimist Playlist

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

Ask about getting content back out of AI, I keep hearing,

"AI is a black box. NO one understands."

And...you accept that. After all, I’m not an engineer.

Stuck in this AI Black Box myth - and told I wouldn't ever understand. Until a listener shared a comment that made me triple take. Dang.

Now with the AI tools to detect and protect copyright, it's time to turn around and stop looking in the black box.

Wow, crack 1 – the AI black box hype may just be that, hype.

Some companies are finding their content inside an AI system with just a few tweaks - content tech companies say is "transformed beyond recognition" and impossible to track.

I've been repeating what the big AI companies told us - once your content turns into "tokens," it's like mixing paint.

You can't unmix it, you can't identify it, you can't remove it.

I bought it. All of it.

And why wouldn't I?

These companies kept saying the same thing - it's all math and statistics, it's all transformed, it's fair use. The black box defense.

Then reality hit me.

The New York Times, in their legal filing, shows how ChatGPT spit out their articles almost word-for-word, possibly by tweaking a "randomizer" setting.

It's like finding a knob on your blender that suddenly unblends your smoothie back into whole fruits.

I found out about Patronus AI and their CopyrightCatcher. They ran a test with ChatGPT and found that 44% of what it produced was copyable content.

Not "inspired by" or "in the style of" - actual copyable content that could land someone in legal hot water.

If AI isn't really a black box after all, all those company valuations built on the idea of "we transformed it, so we own it" come crashing down.

If your copyrighted content (or something close enough) can be found inside these systems, the whole black box story changes.

Now creators gain control and protect their work if they take a few steps.

No engineering degree required. Often for free.

Recreating Copyright Content: Does AI Remember?

The idea that AI as an impenetrable black box? The evidence is hard to ignore:

* Patronus AI's Discovery: 44% of ChatGPT outputs were copyable content in their early test (2024)

* The NY Times Prompt Hack: How adjusting a simple "randomizer" setting revealed much of the original articles.

* The Visual Evidence: Researchers Gary Marcus and Reid Southen showed Midjourney could create images displaying copyrighted characters from Marvel, Star Wars, Sonic, The Simpsons, and other major franchises—proving the same issue exists across text AND visual AI

* The Key Insight: Thomson Reuters, via its Westlaw legal business, won against a Fair Use Defense by Ross Intelligence, who legally purchased data from a 3rd party to train their AI.

* Ross are finding it and removing it from their AI. Copyright fair use defeat.

Turn down the randomness, and out comes content that looks an awful lot like what went in.

This matters because it may mean your work isn't being "transformed beyond recognition" as we've been told—it's being stored in patterns that can be pulled back out. At least in these examples.

Patronus AI's approach worked because they focused on outputs rather than trying to peer inside the box.

Instead of trying to catalog all copyrighted material in training data (impossible), these tools focus on clear ownership of specific works and detecting when those works appear in AI outputs.

It's a bit like how Shazam works. Shazam doesn't need to understand music theory or know every song ever recorded.

It creates unique fingerprints of songs and matches them against what it hears. The same principle applies here - create a unique fingerprint of your creative work, then search for matches.

So if these systems remember and reproduce specific content, shouldn't creators have some say in how their work is used?

That's what the next generation of tools is helping to address.

3 AI Copyright Tools for Creators

You don't need to be a tech expert to protect your creative work in the AI era. A new wave of simple tools is making copyright protection accessible, regardless of tech background.

* Pixsy: How photographers and visual artists can track and protect their work

* CreatedByHumans.ai: A simple way for writers to timestamp their unique style

* Credtent.org: The multi-media platform that takes minutes to register your work

* The Process: Upload → Set preferences → Get verification → Monitor → Take action

All three of these tools help establish clear ownership of your work, set preferences for usage, and join networks of creators with similar concerns.

While individual creators have limited leverage, emerging networks like these may negotiate effectively—just like ASCAP for musicians.

What's powerful about this approach is that you're preparing for the future now.

With California's DELETE Act and the EU AI Act pushing for more transparency, having your creative work properly documented and protected puts you ahead of the curve.

It's not about stopping all use of your work—it's about having a say in how it's used and getting compensated if possible

This isn't about being anti-AI.

It's about finding a balanced approach where AI advances while respecting the rights of creators. The tools exist.

They're generally affordable—often free.

They're easy to use.

For creators looking to protect their work and potentially benefit from AI usage, they're worth exploring.

"There is a crack in everything, that's how the light gets in." Leonard Cohen

Cracks are appearing in AI's Black Box, revealing copyrightable material just as Patronus AI, the New York Times, and Gary Marcus found.

Two distinct cases emerging:

* The "input" case—alleging that OpenAI illegally hoovered up New York Times articles to train ChatGPT without compensation.

* And the "output" case—arguing that when prompted right, ChatGPT may spit out a New York Times article that readers would pay a subscription for.

Why does this matter? Three simple reasons:

First, it makes copyright claims stronger. If AI spits back similar, specific works, the whole "we're just learning patterns" defense falls apart.

Second, we need to watch what goes in AND what comes out of AI. Most tools focus on spotting AI-generated content.

Knowing when your content was used for training in the first place might be more important.

Third, it shows these companies can track specific content if they want to. And if they can track it, they can credit it and pay for it. The technology exists.

Some AI companies are already building systems that tell you where information came from.

When you ask a question, you get both an answer AND the sources it used. Kind of like footnotes in a book, but for AI.

AI isn't a magical black box. It's more like a really advanced library with a randomizer switch.

Sometimes it mixes books together to create new stories, but it can also just read you pages from specific books it remembers.

The "we don't know what's in there" excuse is running out of steam. These systems remember, they store, and with the right settings, they reproduce.

Because creativity isn't contained in black boxes, nor is AI. It's more open than we think, less powerful than we believe, and...

Hungry for human creativity because without it, the data becomes stale.

The output becomes another copy of another copy, copied from someone.

Want an AI that works for you?

These tools are just the beginning. Add to it an audience you build over the next 2-5 years, and you've got something AI can't replicate:

🚀 direct connection with people who value what only you can create.

Protection is becoming easier. Compensation is tricky, but now you've got proof of what's yours. And that proof is the first step toward everything that follows.

AI is more beginning than ending, and despite the headlines, it’s still super early.

And we're beginning to discover what it really means to be human, and the value of creative life in a world where we now know—the black box has cracks after all.

The light getting in? Human creativity more dynamic than AI.

RESOURCES

How a New York Times copyright lawsuit against OpenAI

Generative AI Has a Visual Plagiarism Problem

HarperCollins Faces Backlash Over $2,500 AI Training Deal with Authors

Penguin Random House protects its books from AI training use

Penguin Random House underscores copyright protection in AI rebuff

Thomson Reuters wins AI copyright 'fair use' ruling against one-time competitor

California Delete Act

Authors strike back against AI stealing their books as licensing startup

How does Shazam work? Music Recognition Algorithms, Fingerprinting, and Processing

AI Copyright Tools

Pixsy - Fight Image Theft (founded 2014)

Created by Humans: Jan 2025

Patronus AI (2024)

CopyrightCatcher by Patronus

MUSIC

Open Music Initiative - using APIs

SongSecure

CONTENT AND ART

Credtent.org

Artrights

Verisart - Protect your Artwork

Leonard Cohen photo

By Rama - Commons file, CC BY-SA 2.0 fr

https://commons.wikimedia.org/w/index.php?curid=53046764

AI Optimist Playlist

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

What's Immortal, Always Hungry, and Sucking the Life Out of Creativity?

I’ve been swimming with tech fear cycles for years. The internet destroys privacy (it did). Mobile phones make us antisocial – yup. Social media may wreck democracy maybe.

Each wave brings fear, adapting, and moving forward. But something about AI fears feels different.

"AI is a vampire," a friend complained over coffee.

"It lives forever. It consumes everything in its way. And that's the source of its immortality."

His words hit harder than my dark black coffee. Because it connects dots I've been staring at for months.

People fear AI, like a digital vampire. It lives forever. It owns what it consumes. It turns our work into its power source.

And this isn't the first time we've feared something feeding on human…ideas.

The links between vampire stories and AI fears go deeper than tech worry. They touch on basic questions.

* Who owns what AI creates?

* Who gets credit when ideas change form?

* What happens when creations outlive creators?

When we say AI "feeds" on data instead of "learns" from it, we're using vampire terms.

When we worry about AI outlasting us or changing our work beyond recognition, we're telling scary stories with tech words.

And this feeling that something is taking our work without asking? That's as old as the stories themselves.

AI isn't scary because it's alien. It's scary because it plays two roles at once.

It's both our explorer into new territory and the guardian at the gate testing if we're ready to move forward.

AI makes us face ourselves by creating something that's like us but not us.

The vampire comparison makes sense.

Both vampires and AI show us versions of ourselves we don't like seeing. Both have origin stories about disputed ownership and mixed-up credit.

WAIT, vampires aren’t real?When’s the last time you had coffee with AI…..

Let's go back to the coldest winter Europe ever faced. It's 1816, at Villa Diodati in Switzerland.

This is where both Frankenstein and the modern vampire were born during a storm of creativity – and disputed authorship.

When Lord Byron Didn't Write "The Vampyre" (Got the Credit but not the money)

In 1819, a vampire was born twice. First as a character in a story... and then again when a publisher stole that story and put someone else's name on it.

The story was "The Vampyre" – the first aristocratic vampire in English literature and the grandfather of Dracula.

The book, written by Dr. John Polidori from Byron’s initial idea, had Lord Byron's name on it, but Byron didn't write it.

His friend and physician Polidori did. Neither knew about the publication until it was already circulating.

One story, wrong name, massive influence.

It all began three years earlier during what's known as "the year without a summer."

In 1816, a volcano eruption in Indonesia had blanketed the world in ash, creating a cold, dark summer across Europe.

Think of it as the pandemic of 1816, except caused by a volcano they didn't know about.

Trapped indoors at Villa Diodati in Switzerland, Lord Byron, Percy Shelley, Mary Godwin (later Mary Shelley), and Byron's doctor John Polidori grew bored of reading German ghost stories to each other.

Remember, they had nothing else but books. So Byron proposed a contest – each person write their own horror story.

Mary Godwin went on to create "Frankenstein." Polidori took a fragment that Byron had written and abandoned, developing it into "The Vampyre."

They all enjoyed the stories and moved on with their lives. Mary Shelley later published her work to acclaim, then suddenly Polidori's tale got published without him knowing it...

In 1819, publisher Henry Colburn released Polidori's "The Vampyre" under Lord Byron's name.

Byron had fled England due to debts and scandal, making it the perfect time to profit from his celebrity.

The story became an immediate sensation – not because of its content alone, but because readers believed the famous poet had written it.

When Byron discovered this, he was furious, as was Polidori. But what could they do? It had already spread throughout Europe.

The story was out there, transforming into new works, inspiring new creators.

The pattern set in motion that day reaches all the way to our modern vampire tales – and parallels current AI debates.

Like AI outputs being credited to the company that created the model rather than the contributors who made it possible, "The Vampyre" became known by the most marketable name, not the actual creator.

Seventy-eight years after Polidori's story, Bram Stoker picks up the vampire concept and transforms it into something more powerful – "Dracula" – launching a genre that continues through "Twilight," "True Blood," and beyond.

And that's exactly where the AI conversation gets way complex.

From Gothic Horror to Modern AI: Ownership Matters, Fear Rules

At Anime Expo 2024, a Chinese digital artist named Yuumei is selling her vibrant, colorful work.

A buyer makes a purchase, looks closely at the art and says it was made with AI.

The artist denies it. Provides time-lapse videos of her process. The internet erupts.

But it wasn't AI at all. Yuumei had been tracing another artist's work, particularly an artist known as "Ask."

She called it "over referencing." The buyer gets a refund. The posts get deleted.

We're so focused on catching AI, we question all we see. The fear of AI is so pervasive that it's now a default suspicion when seeing impressive work.

We're not just worried about AI taking our work – we're worried about losing the ability to tell what's human and what isn't.

It's not about the tools anymore. It's about trust.

When an accomplished artist gets accused of using AI, it reveals our deeper anxiety.

The line between human and machine creativity is blurring. We can't tell the difference through our own eyes anymore.

Like it was in the early 19th century (what)?

In 1819, a publisher put a famous name on someone else's work to sell more copies. In 2024, we assume impressive work may be machine-generated.

Both involve the same fundamental question: who gets credit?

In the vampire story, publishing was new, open, with few rules.

That's exactly where we are with AI today.

No clear standards. Uncertain credit. Creative works flowing freely across boundaries we haven't fully defined yet.

And just like the literary scandal that led to Dracula, we're only at the beginning of this.

What matters isn't stopping the change – it's understanding what's driving our fears about it.

Because what scares us about AI, like vampires, isn't just what it does. It's what it represents about us.

The Insatiable Hunger for AI Data

Vampires need blood. AI needs data. Both feast on human creation to sustain themselves.

Both entities transform what they consume into something new – energy, power, capability.

Both raise questions about permission versus feeding.

Both trigger primal fears about something taking what's ours without permission.

So is training AI like feeding, or is it like learning? This question is at the heart of our copyright debates.

If AI is "feeding" like a vampire, then it's taking without giving back to society it drains.

If it's "learning," that's where copyright law arguably shouldn't apply, at least according to countries like Japan. The US isn't so sure yet.

We're in the early days of figuring out what copyright law means in every country. And is the knee-jerk reaction to say "it's feeding on what we have" just a way to stop it?

Or is it a normal human impulse in the face of things that resemble vampires – unknowable, powerful, and transformative?

This hunger for data ownership becomes especially important with AI because we can't unwind it.

We can't go backward. The data is in there.

So, ownership now means that AI companies claim it as transformed content they've learned from.

It’s not there as the content that went in as you remember. As we deal with ownership, we must deal with the transformative nature of what's happening here.

If Bram Stoker had to pay for the inspiration that led to "Dracula," who would have gotten the money?

The publisher who stole Polidori's work? Polidori himself? Byron for his original fragment?

That's the tangled creative “who’s the author” mess we're trying to sort out with AI.

People using ChatGPT for term papers aren't thinking about whether Shakespeare's estate got paid for all those sonnets in the training data.

They're just trying to finish their homework. But that's the thing about AI – we're all using it, but few understand how it works.

When does AI stop being a fancy copying machine and start being... intelligent?

And who gets to decide? Because right now, AI-created work is treated as guilty until proven human.

And that brings us to the copyright paradox itself.

The Copyright Paradox

You know, both Frankenstein and Dracula could get copyrights if they existed today - they used to be human.

And AI? It used to be human too, in the data and code that evolved it. Our thoughts are their ancestors, our dreams its guidance, and our egos its threat.

In the US copyright system, it's like a three-legged stool - take away any leg, and the whole thing falls. Let's see how AI handles each one.

First leg: fixed in tangible form. This is the easy part. AI passes this test without breaking a sweat - generated text saved to files, art appearing on screens. The output exists somewhere real.

Second leg: original work. AI struggles because it's trained on existing works. It can't prove originality. There's no clear chain of creation. Remember The Vampyre? At least we knew where that came from.

Third leg: created by a human author. In U.S. law, an "author" means a human author. Not a machine. Not an AI. A person.

This is why the U.S. Copyright Office won't register purely AI-generated work. It's not about quality. It's not about creativity.

It's about humanity. Even if an AI writes the next "Frankenstein," without human creative input, it goes straight to the public domain.

Think about The Vampyre. Wrong author name? Still copyrightable.

Borrowed heavily from Byron's fragment? Still copyrightable.

Why? Because Polidori was human, and he made creative choices. The publisher might have given false credit, but the human creativity was there.

With AI? We're still figuring out what counts as human creativity. Is a prompt enough?

How much editing do you need to do?

When does AI assistance become AI creation?

Dracula came out 78 years after The Vampyre.

Imagine if Bram Stoker had to track down that publisher who stole everyone's credit and pay them for "inspiration rights."

That's like paying a thief for stealing something that inspired you to make something better.

The copyright paradox creates what some call "the liar's dividend" - the incentive to falsely claim human authorship since AI work can't be copyrighted. And it still hard to prove a fake.

This creates a dynamic where honesty about using AI tools comes with a legal penalty.

And this is where our mirror problem begins.

The AI Mirror Problem

There's an old myth that vampires can't see a reflection in mirrors. Maybe that's AI's problem too - it creates, but it can't see itself creating. It can't understand its own process.

But neither can we, really. Lady Gaga wrote "Just Dance" in 10 minutes. Some artists labor for years on a single work.

When did time become the measure of creativity?

When did proving we're human become necessary to own what we make?

We've created AI in our image, gave it powers that both fascinate and terrify us, and now we're shocked when it reflects our nature back at us.

Classic human move, really. We've been creating external indicators of our internal powers and fears since we painted on cave walls.

What Joseph Campbell, the famous mythologist, would have called the shadow self, projected onto reality.

That's a lot of what's going on with AI.

The Yuumei case shows this mirror effect in action. Here's someone well-known, very successful.

You instantly respect her work when you look at it.

And they accuse her of using AI because AI becomes their fear. "I'm being cheated."

Everything we look at now, we wonder, is that real?

And that could be a good thing. For all the people pushing critical thinking with AI, it might be one of the best outcomes.

People question what they see.

But the more famous you become; the more likely people will accuse you of using AI.

And why is someone accusing AI like accusing them of doing something wrong?

In Campbell's framework, AI plays two mythological roles at once - it's both the hero and the threshold guardian.

It's the adventurer we send into unknown territory, and it's the intimidating figure at the boundary testing whether we're ready to go past the threshold.

What makes AI different isn't that it's some alien intelligence - it's that it's the hero and the threshold guardian in our story.

It challenges us to understand ourselves better by creating something that mimics us yet remains fundamentally other.

Maybe what we fear about AI isn't that it will drain us dry - vampires tend to keep their victims alive, after all.

What we fear is that it reflects something uncomfortably true about ourselves: our endless hunger for more, dreams of living forever, and consuming without consequences.

Maybe what we fear isn't the AI vampire at all. Maybe what terrifies us is staring into that AI mirror and seeing nothing staring back.

This technology holds up a mirror to humanity – one that shows our creativity isn't quite the magical, mysterious process we claim.

It reveals how we learn, how we create, how we transform ideas into something new. And sometimes reflections make us uncomfortable.

When we look at AI-generated work, we're not seeing a monster. We're seeing ourselves – our patterns, our inspirations, our collective creative consciousness. But with the creator's face erased.

I'm not saying ignore the real issues. When your creative work gets sucked into training data without your say-so, that stings.

When companies profit from models built on millions of unpaid contributions, that's unfair.

But painting AI as a blood-sucking monster misses the point. The path from Byron's fragment to Polidori's "The Vampyre" to Stoker's "Dracula" wasn't a feeding frenzy.

It was ideas evolving, transforming, improving – exactly what happens in AI, just faster and at scale.

This is human in the data and the people that created it.

It's human in the thoughts and stories and processes and code that was done by generations of people.

That's why the ownership question is so tricky.

We want clear lines between "mine" and "yours," but creativity has always been messier than that.

AI makes this messiness impossible to ignore.

Remember Yuumei? Her art was human-made but she got accused of being AI-generated.

The border between human and machine creation is blurring, not because machines are becoming more human, but because we're finally seeing how pattern-based human creativity really is.

And maybe we don't need to fight the vampire. Maybe we need to accept that our ideas live beyond us – transformed, reimagined, built upon.

That's not immortality stolen by a monster. That's the immortality we've always had.

As Joseph Campbell might say, AI isn't our enemy – it's our technological shadow self, reflecting both our strengths and deep flaws.

We've projected our consciousness onto code, just like ancient people projected it onto cave walls.

Neither vampires nor AI as a vampire is real.

It's a technology and intelligence and a way that we're looking at ourselves that sometimes we don't like, and other times is the key to understanding what we can become.

The mirror may be blank, but what do we do with that reflection?

That's entirely up to us.

Resources

Yuumei Art

The AI Vampire at the Door

Interview With The AI Vampire, Part 1

Villa Diodati

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

In Tokyo right now, an AI can legally train on nearly everything created in Japan - books, songs, artwork - without permission or payment.

This isn't science fiction; it's current Japanese copyright law.

Step outside Japan's borders, and suddenly those same works may gain legal protection. This isn't theoretical - it's playing out in courtrooms today.

The story of Ultraman - one of Japan's superheroes - represents this contradiction. In a Chinese courtroom, this Japanese character won a copyright battle against AI.

Yet back home in Japan, that same AI use – for training - would be completely legal.

This isn't about superheroes and science fiction.

It's about your work becoming someone else's fuel, with different rules depending on which digital border you cross.

Because Japan’s AI and China’s AI and the US’s AI….countries will probably have their own, as well as their own copyright rules.

In 2025, we're witnessing the emergence of AI zones - digital territories where copyright means different things in different places.

There won't be one global system, it’s a patchwork of nationalistic and cultural points of view.

What Japan's model reveals isn't just a legal curiosity - it's a potential future we're all heading toward, where content flows freely into AI within certain boundaries, but with protection against direct copying or financial harm.

When your work can travel anywhere but the rules protecting it cannot, you need to understand the global rules - not just your local square.

The AI Wrapper Not the Site: What Creators Learn from Ultraman

When you're fighting AI copyright violations, you need to know where your content is located. Except now it’s all tokens and math.

Only thing to do is try to recreate what went in….

A Chinese licensing company (SCLA) owns Ultraman rights in China. They found their character being generated by an AI system.

They took legal action against the website where users could enter "Ultraman DYNA" and get images that clearly resembled their licensed character.

They won - but they didn't actually solve the problem.

Because they targeted the website, not the AI system feeding it. Where the Ultraman imagery still resides.

This matters for creators because:

* Many AI websites are "wrappers" - they don't own or train the AI models

* They're paying monthly fees to connect their site to third-party AI via APIs

* The Ultraman images are still in the AI's training data, even after the lawsuit

* The court couldn't remove what wasn't part of the case

When you use a site with AI features, chances are they're just an interface connecting to models built elsewhere (obvious for people in AI, surprisingly not for many who are not familiar).

For creators standing up for their work, this means tracking the data to its source.

Don't just go after the visible website - follow where your tool gets its training data. Otherwise, you'll win the small battle and your content will still be able to be reproduced.

And as mentioned everywhere, no one is sure they could actually pull your content out of AI, because your content isn’t your content anymore.

Do AI Tools Really Need Artists?

Witness Japan's approach to copyright and you might wonder if AI needs any help. Their system allows AI to train on copyrighted works, but with guardrails that might surprise you.

AI working with creators, like software developers have for years.

* Adobe Firefly, trained on licensed stock, gains widespread adoption among professionals

* Sony's Flow Machines co-creates with artists, sidestepping legal battles

* Midjourney asks users for feedback and allows creation of mood boards to improve its systems

I'm not even close to being good on Midjourney, but it ups my game tremendously by providing tools smarter than I am as a novice graphic designer.

That's not replacement - that's enhancement. Always asking for feedback like a strong AI startup.

Despite drawing a line in the sand with copyright, that same line may bring us together.

The Japanese experiment recognizes a fundamental truth:

AI gets better when humans remain in the loop not just as data sources, but as active participants.

The question isn't whether AI needs artists - it's how we structure that relationship to benefit both.

As much as we might want to keep separate worlds, the evidence shows this is where AI gets good - when it learns with us, not just from us.

2 Parts of Japan's AI Copyright rules

The Blueprint Behind Japan's AI Freedom

What makes these AI tools work isn't just tech brilliance - it's the legal framework supporting them. Looking at Japan's approach reveals something engineers have been arguing for: a system where AI learns from everything while still respecting creators' rights.

Japan's copyright rules for AI rest on two key pillars that balance progress with protection:

The Non-Enjoyment Purpose Requirement

The first pillar is simple: intention matters. Japan allows copyrighted content to train AI systems if the purpose isn't to recreate the original work.

You're not bringing in Van Gogh paintings to create more Van Goghs - you're bringing them in to help the AI understand artistic styles, composition, and color theory.

This creates breathing room for AI development while maintaining a boundary around blatant copying. The focus isn't on inputs but outcomes.

While this sounds straightforward, it creates real questions for creators: how do you distinguish between "learning from" and "copying" in a practical sense? This leads to Japan's second pillar.

Article 34 Proviso: Where Protection Lives

Japan doesn't leave creators without protection. Their Article 34 Proviso acts as a safety net, establishing that if AI output causes actual harm to a creator's career, reputation, or financial status, normal copyright protections kick in.

This practical approach says: let data flow into AI systems, but watch carefully what comes out. If the output creates direct competition or damages creators, that's where the line gets drawn.

Fair Use and Japan's AI Freedom - not the same

How different Japan's approach is from the U.S. "fair use" defense - both trying to solve the same problem.

In the U.S., fair use isn't a rule - it's a legal defense. You only know if something is fair use after you've been sued, and a court decides. It's reactive, not proactive.

In Japan, they've built a system that's more predictable. AI companies know upfront what's permitted: training is allowed, copying isn't.

Creators know where their protection lies: not in restricting input, but in preventing harmful output.

This is what many U.S. engineers are actually arguing for when they claim fair use. They're saying:

"We're not taking content to reproduce it - we're taking it to learn patterns, just like Japan allows."

Both agree, what comes out matters more than what goes in. If the output looks like what went in, both countries consider that a violation.

As we move through 2025, the U.S. Copyright Office findings will shape everything.

I suspect we'll see something like Japan's system emerge - allowing inputs for learning and focusing protection on outputs causing real harm.

The "Human-AI Collaboration" Spectrum

I'm amazed the US Copyright Office didn't simply declare all content suitable for AI training. By leaving it open, we're in this year-long limbo - but I can't imagine they'll crush ChatGPT, Claude, and the entire AI industry with a finding against it.

What we're likely to see instead is a spectrum of human-AI collaboration, with different levels of protection based on human involvement.

This isn't just speculation - it's already happening in courts across the world, including Japan and China.

At one end of the spectrum, you have minimal input - dropping in a prompt or two and accepting whatever comes out.

Courts are already signaling this won't qualify for copyright protection.

On the other end, iterative creation - refining, directing, and shaping the AI's output through rounds of feedback.

The courts are looking for evidence of your creative process - not just that you typed "Ultraman DYNA" and grabbed the first image.

They want to see focused human effort with the tool, making deliberate choices, and guiding it toward your vision.

What's emerging isn't binary (all human vs. all AI) but where protection increases with human involvement.

The U.S. Copyright Office is likely heading toward a similar model - not copying Japan, it reflects the reality of how we're both using these tools.

Concept and Direction? Prove it!

So what does meaningful human input actually look like in practice? Two words are becoming crucial: concept and direction.

Courts are looking for evidence that you had a concept in mind before approaching AI. Did you have an intention, a vision, a goal?

Or were you just fishing for whatever the AI might generate?

This matters because intention shows you're bringing something original to the table - not just activating a random content generator.

Direction is equally important - can you demonstrate how you guided the AI toward your vision?

This means showing your creative choices - whether that's choosing specific platforms, curating between different AI outputs, adjusting colors, angles, or filters.

You need to show more than just button-pressing.

I've noticed the U.S. Copyright Office repeatedly questioning whether works were "just done by AI" - and without evidence of human direction, they're likely to deny protection.

The more you can document your process - showing your inputs, revisions, selections, and edits - the stronger your claim to copyright protection becomes.

The real question isn't whether AI was involved - it's whether a human mind was driving the creative process rather than following it.

No Easy Answers for AI - finding your way

Just like that Chinese court in the Ultraman case, there are no easy answers here.

While they found for the license owner, it was a minimal fine that didn't actually stop the problem.

Not because the court was wrong, but because the plaintiff didn't understand where the real infringement occurred - in the AI's training data, not just on the website.

Most people listening right now probably don't understand the technical architecture I just described - how websites are often just interfaces connecting to third-party AI systems.

This disconnect between what we see and how things actually work creates problems for copyright enforcement.

So where does that leave creators? Four principles can guide you forward:

First, understand that there's no going back. The genie is out. AI has seen the work of humanity and will continue to learn from it.

Second, document everything. If you're creating with AI assistance, maintain records of your process. Save your prompts, iterations, editing decisions, and selection criteria. This evidence of your creative direction will become increasingly valuable.

Third, focus on curation and assembly. Even as individual elements become easier to generate, the thoughtful selection and organization of those elements remains distinctly human.

Copyright protection for compilations and collections is well-established, even when the pieces themselves might not qualify.

Finally, embrace the spectrum. The future isn't binary - AI or human. It's finding the sweet spot where technology amplifies your creativity rather than replacing it.

The most successful creators won't be those who avoid AI entirely, nor those who rely on it completely, but those who learn to dance with it.

As much as we've drawn copyright as a line in the sand, it can bring us together - forcing creators and engineers to find mutual solutions rather than mutual destruction.

The Japanese experiment offers a glimpse of what's possible:

a system that acknowledges both our collective interest in advancing AI and our individual interests in protecting creative work.

No court decision will provide perfect clarity.

No copyright ruling will solve every problem.

By focusing on what we can control - our own creative processes, documentation, and adapting - while continuing to create work that matters.Resources

Chinese Court Issues First Decision on AI Copyright Infringement

China’s First Case on AIGC Output Infringement--Ultraman

28 February 2024

Report on AI and Copyright Issues by Japanese Government

NO&T IP Law Update

27th_Tokyo_International_Film_Festival_Ultraman_from_Ultraseven_(15001540863)

By Dick Thomas Johnson from Tokyo, Japan - 27th Tokyo International Film Festival: Ultraman from Ultraseven, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=118426646

By Dick Thomas Johnson from Tokyo, Japan - Ultraseven from "ULTRASEVEN 55th Anniversary Special Screening" at Red Carpet of the Tokyo International Film Festival 2022, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=124731453

SCLA

What is an API wrapper?

Google's Project Magenta Brings Neural Networks to Music

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

Five years ago this Valentine's Day, OpenAI releases GPT-2; headlines later warn it’s 'too dangerous to release.'

Their concern? That it could generate synthetic content at scale.

Fast forward to 2025, and we're seeing the first major AI fair use victory in court - Thomson Reuters just won an early judgement protecting their platform's content from training a competitor’s AI, something courts rarely find copyrightable.

The other side bought the data from a third party and used it to train its AI, claiming fair use. And in this round, that argument was lost, with more details later in this article.

The fair use claim is the same one used by ChatGPT and most others; while the facts are quite different, the fair use denial is an interesting response to turning content into tokens. In this round at least.

Roulette Wheels or emerging Intelligence?

Last week we looked The U.S. Copyright Office releasing their second report stating AI-generated work isn't copyrightable because it's essentially a 'roulette wheel of probability.

While everyone's focused on AI outputs, the real Achilles heel weakness is hiding in plain sight – what do we do with content used to train AI without permission, AND not falling under fair use?

It's about the millions of novels, articles, paintings, photos, and lines of code that built these AI systems.

Next year the US Copyright office will release the 3rd update, this one on what happens when content goes into AI, from copyright law in the US. Surprising that the first 2 reports depend on this 3rd one, also showing how complex and new this issue is to all of us.

How can you issue AI a copyright when you can’t define what part of someone else’s content went into what came out?

What if they found all the data in all the AI is not fair use; it’s all copyrightable content that should have been given something – an email, a notice, something. Or not.

There's no way the Copyright Office would risk shutting down ChatGPT and Claude by extending protection now. It will be 6 years since this began when the update should come out, and the findings will be hard to use retroactively.

But what happens when they finally tackle the question of all that creative work that went into building them?

Like the Thomson Reuters winning copyright protection for it’s Content last week!

Ross, a startup, competes with Westlaw's legal research platform. And had worked there previously, so likely knew the quality of the information for training AI would be high.

They needed legal Q&As to train their AI - think of it like needing textbooks to teach a student.

When Thomson Reuters said 'no' to licensing their content, Ross got creative... maybe too creative.

They bought data from a company called LegalEase to get something called 'Bulk Memos' - basically, lawyers' compilations of legal questions with good and bad answers.

These lawyers were using Westlaw's headnotes (like Cliff Notes for legal cases) to create these answers. Ross ended up with 25,000 of these memos to train their AI.

This is eerily like how OpenAI, Anthropic, and others acquired their training data. They purchased crawled data from Common Crawl - think of it as a digital library that copied everything on the internet.

Like Ross tried to sidestep Thomson Reuters by getting content through a third party, AI companies have been using third-party scraped data without creator permission.

Businesses like Common Crawl and even WayBackMachine.org have been collecting this data, but both are non profits. Like OpenAI…wait, OpenAI is going away from that now. Back then?

It’s never been a big deal until AI companies make billions with that data, at a scale way beyond anything before.

The scale is astronomical. We're not talking about 25,000 memos - we're talking about trillions of data points from millions of creators.

While trillion-dollar tech companies claim they can't pay for content, they're quietly making deals with publishers who have the legal and financial muscle to fight back.

The New York Times, Axel Springer, even Thomson Reuters - they're all getting seats at the table.

But what about the millions of independent creators whose work built these AI systems?

And that brings us back to 1787, when our founding fathers picked up their quill pens...

To know where copyright is going, you need to know where it starts…in the US for our example.

The origin story of AI Copyright….really all US copyright, begins with that document and a promise to protect inventors and creators.

Smart as he was, I don’t think Thomas Jefferson could wrap his head around our questions….

Constitution Origins: From Quill Pens to Neural Networks

In 1787, the framers of the Constitution faced a fundamental challenge: protecting creativity in their new nation.

Article 1, Section 8 states: "To promote the progress of science and useful arts by securing for limited times to authors and inventors the exclusive rights for their respective writings and discoveries."

The Copyright Office maintains that purely AI-created works don't promote human progress because they lack human creative input.

But in an era where AI systems can solve complex scientific problems and create new art forms, we're redefining what 'progress' means.

"See, when the Constitution talked about respective writings and discoveries, right? It's like back in the day when they didn't have anything but writing, photography was coming to really blow their minds very soon."

The Sui Generis Framework: A New Model for AI Copyright

"Now, the reason I'm proposing the sui generis option the US Copyright office didn’t pursue, is a customized legal view of AI generated work is because right now, there's agencies and people doing work with commissions being told and instructed what to do to create images like cases we've shared with you before."

That basic service of delivering copyright protection for a work for hire, or other arrangement, is left out on its own with AI. If AI generates it, it’s not human, and it’s not copyrightable.

Sure we all bend the rules in the meantime. Until a big mistake happens. And the liability goes to the one who paid for the work, yet doesn’t have a copyright to protect themselves.

The framework needs to address four key areas:

1. Authorship and Ownership

"It's also a focus on substantial human involvement. How much? And this is case by case, but how much was this person involved in creating this or how much was the AI involved?"

2. Duration of Protection

"One thing people complain about copyrights in traditional ways is how long they are... And for AI generated work it's clear we're looking like 5 to 10 years."

3. Fair Use Evolution

"Fair use is really important... There's many different ways that people can use it and be able to use the content without taking it too. Certainly, it doesn't mean take the book. You might take a quote."

The Creator's AI Copyright Playbook

"Document any of your work with generative AI. And by the way, with anything, if you're not using AI do the same thing because there's going to be more requirements."

Strategic Documentation

"Make sure you can prove whether you did something with AI or without it, and show a record of your activity. Photos, manuals, journals, whatever. You have calendars to keep proof of what you've done."

Transform, Don't Just Generate

"When AI generates independently, in other words, I put a couple of prompts in, spits it out. I want a copyright. No, because prompts don't give copyright. Not in China, not in the US as of this time."

Building Your Evidence Trail

"Now when you transform AI output, when you take AI output and turn it into something else, that's possible to get a copyright. Assuming you're not using AI to transform AI, right?"

The last US AI Copyright update was so wrong to me, AI being just a 'roulette wheel of probability.'

And in their early update on AI Replicas, they wouldn’t touch the issue of copyright because of likeness and image – and no rules applying there. Not because it was machine generated.

Because it was not human and it’s a word calculator. A spinner of other’s tales. Bringing it all down to the lowest common denominator.

Enough….and looking around at news in the past 10 years, what you see is likely what we’ve been creating, because all that “bad” data….is also us.

It’s a moving target, not a set in stone and time. New copyright is like playing a game, old copyright is like reading a book.

Both are fun, one moves a little faster than the other.

The US Copyright office will likely have a new challenge with its findings on how copyright applies to content put into ChatGPT, and everyone else, by this time next year.

Enter the Humanity's Last Exam – HLE - a test so challenging that no single human, not even Nobel laureates, could ace it all. We're talking 3,000 questions across every field imaginable, from philosophy to quantum physics, developed by nearly 1,000 experts across 500 institutions.

In just one year, AI performance on HLE jumped from 10% to 30% accuracy.

It's not just predicting anymore, it's reasoning through problems that challenge the brightest human minds. It's like watching evolution in fast forward.

And there are many more. The GAIA (General AI Assistants) benchmark is an evaluation tool assessing the performance of AI systems, with a focus on their ability to function as general assistants.

Is AI ready for the big questions – right now it depends how you pose them, and how you train AI, now, 2 weeks from now, and every 2 weeks.

Creators listen: If AI triples its performance again this year - and at this rate, who's betting against that? - we're not talking about a fancy probability engine anymore.

We're talking about something that could redefine human knowledge itself.

Ready and perfect, no way – way more interesting than anything in the past 20 years?

Maybe….and it’s already in our lives, many simply don’t understand how…..

So while the U.S. Copyright Office wrestles with digital replicas in one report and dismisses AI as random in another, the reality is racing ahead of our legal frameworks.

That's exactly why we need your voice in this conversation.

Because the future of creative rights isn't just about law - it's about understanding what happens when machines start thinking like humans... or maybe even better.

When we start thinking with and working with AI, maybe it helps us grow as well.

Share your thoughts in the comments.

How do you think we should protect creativity in a world where AI isn't just rolling dice anymore?

RESOURCES

NOTE: Thomson Reuters owns Westlaw. When I refer to Reuters in this video, it means Thomson Reuters.

Thomson Reuters wins AI copyright 'fair use' ruling against one-time competitor

By Blake Brittain February 11, 2025

THOMSON REUTERS ROSS LAWSUIT fair use.pdf

The Automation Assumption Was Wrong—The Revolution Eats Itself

Humanity's Last Exam (HLE)

Common Crawl

AI deemed ‘too dangerous to release’ makes it out into the world by Andrew Griffin

New York Times sues Microsoft, ChatGPT maker OpenAI over copyright infringement

US copyright office

Part 1: Digital Replicas

Part 2 (out January 2025): Copyrightability

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

What? Jackson Pollock Might Help Us Understand AI Copyright?

I see Jackson Pollock's name in the US Copyright Office's announcement about AI this week, and I'm thinking - how did a 1950s painter who threw paint at canvases end up in a debate about artificial intelligence?

From the US Copyright Office

Someone tried to connect Pollock to AI in a way that made me stop and think: What if Pollock's paint splashes were like our prompts to AI?

We throw words at it, and something unexpected comes back - something we didn't necessarily visualize, and it's still an act of creativity.

The Copyright Office didn't buy it. They looked at Pollock comparison and said,

"Hey, there was a human holding that paintbrush, choosing colors, making decisions about where to throw the paint."

To them, AI is more like a "digital roulette wheel" (their words, not mine) that randomly spits out content based on prompts.

The New York Times Might Just Prove Them Wrong

The Copyright Office gives zero respect to prompts - and I mean zero. They're basically saying prompts are like throwing spaghetti at the wall to see what sticks.

But The New York Times lawsuit against OpenAI might flip this on its head.

"If the New York Times really got their close to original content out... that actually is the first time we've connected this specific action to the specific prompt.

Every prompt shaped an output, just like every Pollock throw of paint created the canvas."

If The Times prove they used specific prompts to get their exact content back from ChatGPT, doesn't that show there's more to prompting than random chance?

Three Acts of the US Copyright Drama

We're in the middle of a three-part series from the Copyright Office:

1. They've covered Part1. Digital Replicas (think deepfakes) in July 2024

2. Now they're tackling AI authorship: January 2025

3. Next year's the big one - what happens when AI trains on copyrighted content?

What They Actually Said

Here's where they got something right: you can use AI in your creative work, you just can't let it do all the heavy lifting.

Take "Zarya of the Dawn" - this cool comic book where the images came from Midjourney, but a human did the writing and put it all together.

The whole book? Copyrightable. The individual AI images? Not so much.

Two Old Cases That Suddenly Matter Again

The Copyright Office pulled out some fascinating historical cases that suddenly feel super relevant:

In 1989 a homeless advocacy group commissioned a sculpture and then argued they owned it because they gave the idea? (CCNV vs. Reid) Sound familiar? It's basically the prompt versus output debate we're having now.

Or the map reproduction case (Andrein) that helped us figure out the difference between "creative guidance" and "just following instructions"? These old cases are helping us navigate this new AI territory.

Maybe AI is Human Intelligence

I think we're looking at this all wrong. When we call it "artificial" intelligence, we're missing something huge. Here's my take:

"Without that data, none of this works.

Without the code created by humans, none of this works.

Without the person prompting or talking... it's all just language communication with a database of incredible collective intelligence."

Maybe our real problem isn't about respecting AI's copyright - it's about recognizing intelligence that goes beyond a single human author.

We're trying to force 2024 technology into 1787 rules, and surprise, surprise, it's not fitting very well.

The Copyright Office got one thing absolutely right - they're protecting AI-assisted human creativity. But they're missing the bigger picture.

AI isn't just some random number generator spinning out content like a Vegas slot machine. It's human intelligence evolving into something new.

If AI is built on human knowledge, trained on human creativity, and guided by human prompts, maybe we need to stop thinking about it as "artificial" at all.

Maybe it's time to recognize it; our collective intelligence taking on a new form.

Next week, we're going to explore a new framework for protecting AI-assisted creativity - one that doesn't try to squeeze it into 18th-century rules or dismiss it as luck.

What we're really talking about isn't artificial intelligence at all - it's human intelligence evolving.

And that's something worth protecting….

RESOURCES

US copyright office

Part 1: Digital Replicas

Part 2 (out January 2025): Copyrightability

Part 3 to come

OpenAI claims New York Times' prompting strategy violates its terms of service

New York Times sues Microsoft, ChatGPT maker OpenAI over copyright infringement by Ryan Browne

OpenAI says New York Times 'hacked' ChatGPT to build copyright lawsuit

Copyright and AI

Community for Creative Non-Violence v. Reid, 490 U.S. 730 (1989)

Thanks for reading The AI Optimist! This post is public so feel free to share it.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

Everyone's talking about DeepSeek building a ChatGPT competitor for $6M.

Maybe it works, maybe it doesn't – we'll see.

But while we're arguing about AI chips and compute power, we're missing the real story:

China is the only country on Earth that's actually protecting AI creativity.

Not Silicon Valley with its billion-dollar legal battles. Not Europe with its complex regulations. Not Japan with its tech prowess. China.

Here's what they've built since 2018: The Beijing Internet Court.

Think small claims court meets digital age. Anyone – individual creator or small company – can submit an AI copyright case online.

No million-dollar legal teams required. No years-long waiting game. Just practical decisions about real creative work.

Want to protect your AI art?

In the U.S., you're looking at a 3-5 year legal battle that costs more than most creators will ever make.

In China? Submit your case online, show your creative process, and get a decision that looks at what you actually did – not what some 100-year-old law says about human authorship.

China isn't just protecting human creators using AI. They're also granting copyrights to AI apps themselves.

Why? Because people created those apps.

"They found that the AI was created by people.

Therefore they are the authors and the owners.

The Chinese legal courts are that far ahead of us in AI, almost five years ahead of us." "

And they're doing it through a system built for the digital age – specialized courts handling online cases, practical decisions based on real creative work, and protection for both individual creators and AI developers.

While Western courts debate whether AI can create at all, China has:

- Built an e-litigation platform anyone can use

- Created specialized courts for digital creativity

- Protected both human creators using AI and AI apps created by humans

- Made practical decisions case by case

- Done it all without massive legal fees

The country we often think limits individual rights is actually protecting individual creators – both those using AI and those building it.

While we're stuck in theoretical debates about human vs. machine creativity, China's building a system that works in the real world.

One company spends $6M, another spends $6B, for essentially the same product.

Just like DeepSeek, copyright has a far cheaper solution, and it's already working in... China (not again!).

Three AI Copyright Stories That Show Us Why

1. Jason Allen's Digital Opera (2022)

Let's talk about what happens when an artist who understands both code and creativity tries to get copyright protection in today's system.

Jason Allen didn't just click a button and get lucky.

He created "Théâtre D'opéra Spatial" through a process that would exhaust most traditional artists:

* 624 precise prompts in Midjourney

* Professional post-processing in Photoshop, applying years of digital art expertise

* Advanced refinements using Gigapixel AI, a tool typically reserved for professional photography work

* Multiple rounds of iteration, adjustment, and fine-tuning

"The man did so many prompts and so much work on this image.

He brought together skills that didn't even exist when copyright law was written:

Can you imagine them even trying to conceive of a creative person like Jason now, who can code, who can do graphic design and UI and deliver it all in a gaming interface?"

The result? First place at the Colorado State Fair's art competition. The judges didn't know it was AI-assisted. They just knew it was good.

Then came the U.S. Copyright Office.

Despite showing:

* Hundreds of documented creative decisions

* Professional-level post-processing work

* Technical expertise across multiple tools

* A clear creative vision from start to finish

"They denied him a copyright simply because there's a line down the middle here.

You are a human or you're not. Nothing non-human gets it."

Allen did exactly what artists have always done – used the best tools available to bring his vision to life.

“If he had done this in China, I'm sure he would have gotten the copyright."

The Copyright Office couldn't even explain exactly what part was "non-human" because the court has no way of knowing what AI is considered, what an output is.

They just knew AI was involved somewhere, and that was enough to say no.

2. Tencent's Breakthrough (2019)

Back in 2019, Tencent faced a human challenge.

Their AI system, Dreamwriter, was creating automated stock market reviews. Standard stuff, right? Not quite.

The Chinese court looked deeper:

- They examined how Tencent's team selected and developed databases

- They analyzed the algorithm development process

- They studied how the team arranged and selected content

The court's conclusion?

"The AI was created by people. Therefore they are the authors and the owners."

This wasn't just about who pushed the buttons - it was about recognizing the human creativity behind the AI itself.

"This stuff can't just happen.

Even though we pretend copyright law, it's only human beings... It has to bring in people."

Spring Breeze Brings Tenderness by Li - SOURCE: Lexology

3. Li v. Liu: The Image That Changed Everything (2023)

November 2023: Li creates an image called "Spring Breeze Brings Tenderness" using Stable Diffusion. But this wasn't just random prompting.

The court examined every detail:

- Platform choice: "The court said that was absolutely an artistic choice"

- 30 specific prompts defining what to create

- 120 negative prompts specifying what to avoid

- Parameter tweaking showing deep technical understanding

- Post-processing refinements

The judges, who admittedly weren't AI experts, recognized something crucial: these were human creative choices.

"These choices he made met the level of 'intellectual creations' unquote, by the Beijing Internet Court."

Recognizing the Change instead of Trying to Move Backward

China's approach isn't perfect, but it's working. While Western courts say "no" and demand deep pockets for lengthy battles, China's Internet Court system (created in 2018) handles these cases quickly, practically, and case-by-case.

"There's a new intelligence.

Maybe it's super, maybe it's artificial, maybe it's collective intelligence, but it's here.

And to deny copyright is to deny its existence."

We need to recognize collective intelligence because it is all human intelligence - to say it's non-human in AI misses the point.

The original creations were all from us. Even if AI makes up stuff, shouldn't we be able to copyright originality?

Why do we want to keep fighting it?

Oh yeah, because all the copyrights are owned by the money and the lawyers that are going to settle this in 5 or 10 years, while the rest of us just sit on the sidelines.

Here's the wake-up call: China's ahead in understanding AI Copyright, but they're also smart.

By allowing these small cases to come in and not deciding them only on the big players, they're building a system that works.

"Hey, everybody, let's copy China in this way.

They are leading, and you don't really need to look too deeply into it.

They're trying to do something different."

Thanks for reading The AI Optimist! This post is public so feel free to share it.

RESOURCES

* China: A landmark court ruling on copyright protection for AI-generated works (Li vs. Liu AI Assisted Art)

* China’s First Case on Copyrightability of AI-Generated Picture

King & Wood Mallesons

* Beijing Internet Court ruling in first case of copyright infringement of AI-generated article

* Beijing Internet Court

* Is the Chinese 'Dreamwriter' Case Really a Groundbreaking Case for AI-Generated Works?

* A shocking Chinese AI advancement called DeepSeek is sending US stocks plunging

* ‘Someone had to be first': Pueblo artist criticized after AI painting wins at Colorado State Fair

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com

Over 20 major news publishers like NewsCorp, the Associated Press, and Vox Media have rushed to sign deals with OpenAI, with Vox calling it a "strategic content and product partnership."

The stakes?

OpenAI is offering between $1-5 million per year to license copyrighted content, with News Corp potentially securing $250 million over five years.

But this isn't just about big media companies protecting their turf – it's about the future of human creativity itself.

The Copyright 2-Step: Who wrote it (author), and who owns it?

When a photographer named Napoleon Sarony went to court in 1884 over a photo of Oscar Wilde, the simple question: could you copyright a photograph when a machine – the camera – did the actual capturing?

The courts said yes, because the photographer made creative choices. The camera was just a tool.

Fast forward to today. I took those same historic photos into Midjourney and asked for exact replicas. The result? The AI turned something unique into something average.

That's not just a technical failure – it's a warning about what happens when we reduce creativity to "content."

Tying old Laws to New Ways - AI Case law simply isn’t relevant

Both the US and EU are wrestling with a fundamental question: what constitutes "sufficient human authorship" in the age of AI?

The US Copyright Office's rejection of copyright for the AI-generated graphic novel "Zarya of the Dawn" highlights this challenge.

While a human arranged the AI-generated images, the Office deemed the images themselves ineligible for protection.

Meanwhile, the EU's approach, with its focus on human intellectual effort, raises its own questions.

* How do we determine ownership when AI assists in creation?

* Could specific provisions for AI-generated works differ from traditional copyright rules?

"The edge cases are where creativity begins. That's where the really new change, things that people don't first get, comes from."

Jason Allen won an award but couldn’t get a copyright because he used AI. With over 600 prompts, detailed editing after that, since AI was involved, no copyright.

"Remember that spectrum of human involvement we talked about? The courts say you need 'sufficient human authorship' - but what does that actually mean when you're working with AI?

On April 9, 2024, Representative Adam Schiff brought The Generative AI Copyright Disclosure Act to the House floor – a bill that would require developers of generative AI technologies to identify the material they have used to train their systems.

Think about Sarah Andersen and her fellow artists. Their entire portfolios - decades of work - were used to train AI without anyone asking.

Now big publishers like Axios are getting paid for their content through partnerships, while individual creators are left wondering: 'Hey, what about us?'

Schiff’s Act is under the illusion that it’s simply a request to find out what’s been put in, and then identifying ways to extract it which is borderline sad perception of how this works.

The copyright world doesn’t recognize AI as anything, not even officially a tool, and we’re spending our time trying to prove there’s more to copyright than humans, or human intelligence.

* The "Human Authorship" Hurdle: US copyright law requires human authorship. No idea what to do when AI tools are heavily involved in the creative process.

* Case in Point: The Copyright Office's rejection of copyright for the AI-generated graphic novel "Zarya of the Dawn" highlights this issue. While a human arranged the AI-generated images, the Office deemed the images themselves ineligible for protection.

* Defining "Sufficiently Creative" Human Input: How much human input is needed to transform AI-generated content into a copyrightable work? There's no clear answer.

* Gray Area: If a user provides detailed prompts and iteratively refines AI output, is that enough to claim authorship? What if they use AI to generate variations and select their favorite?

* Jason Allen, a Colorado artist, won first place in the digital arts category at the Colorado State Fair in 2022 with his AI-generated artwork titled “Théâtre D’opéra Spatial.”

* Potential Solutions:

* Thresholds of Originality: Establishing guidelines on the level of human creativity required for copyright protection in AI-assisted works.

* "Human-AI Collaboration" Category: Creating a new category of copyright for works where humans and AI contribute significantly, with ownership shared or determined by specific rules.

Time is the hidden factor here. These creators spent years perfecting their craft, building their portfolios, engaging their communities. Meanwhile, AI can generate similar-looking work in seconds.

But here's the fascinating part - those quick AI generations? They're built on the backs of all that human time and creativity. The publishers figured this out first - that's why they're getting paid. Eighteen major deals and counting.

But what about Mumsnet's community members? What about indie creators? What about you? When you collaborate with AI, you're not just investing your current time - you're building on everyone else's past work too. So how do we make this fair for everyone?"

Three Paths Forward

1. Shakespeare in Code

King Lear wasn't performed much until the 1800s – two hundred years after it was written. It was noise back then, until it broke through later.

Society doesn’t always understand what’s really good quickly, and in AI those are called edge cases. If we bury the world in what’s probable, we miss what’s on the edge.

Use influence vectors to find what’s often underneath the gaze of the public, and let that edge influence change in ways it does in human life, profoundly.

Incentivize the edge, pay them to create and invent and make your LLMs smarter. It won’t take an engineer’s salary, and performance driven content should be rewarded in pay, use of technology, and compute credits.

2. The Open Canvas Approach

Human creativity is open source. Instead of the "delay, bluff, and pay" strategy we're seeing with big publishers, imagine a creative ecosystem where contribution means compensation – not just in money, but in access, technology, and opportunity.

Make it like the Axios-OpenAI deal: use of technology in exchange for permission to ingest and reference content.

Plug a culture into AI, creativity arises. Don’t try to tamp it down with lawyers.

3. Think Different, Not Like Me

"We're building AI from what's already popular, processed, proven. And we're risking missing what's coming, what we're inventing, what nobody even gets except somebody who's very young and they see things that you and I don't understand."

And like every other social network, with enough volume of activity, there’s a reward, maybe a payment, maybe a discount on using AI tools.

Grant and scholarship programs to reward human ingenuity that only makes models smarter.

Algos may amplify sameness, AI needs to watch for the subtle differences, the tones that drive humanity. It’s not just about being the smartest, it’s about being aware together.

Like open source, the shared knowledge is developed into something better, with human effort.

Because without human effort, at least in the short term, you’re not going to get great things out of AI.

The Human Stakes

This isn't about creators versus engineers – it's about human creativity itself. The same AI systems training on artists' work are learning from countless GitHub repositories, turning years of engineering expertise into automated code completion.

Entry-level programmers and mid-career developers are facing the same questions as writers and artists: what happens when your craft becomes training data?

And where is all the new engineering talent going to learn if there are no entry or mid-level jobs?

No one's asking for special treatment. Engineers who built vibrant open-source communities, artists who shared work under Creative Commons – they didn't expect to get rich.

They expected respect, permission, attribution – the basic dignity of choice.

The Creative Edge

The truth about creativity: if you don't encourage the edge – the people on the fringe who don't see, don't do, don't act like everyone else – to express themselves, you lose more than just their voice.

You lose the edge itself.

You lose what shows us what could be, not just what was or what's popular now.

Even in a world of endless content, Netflix still needs that one wildly different show that breaks all the rules.

Because that's where the new comes from. That's where we learn what's possible.

While AI frees up time, defining what we'll all do with this time is more important than ever.

For the few who choose to be creative, it's crucial to nourish and nurture, not simply take what they create without reward.

Think AI's amazing?

There's someone out there young and seeing things differently who will open our eyes and hearts, and that experience will make AI even more intelligent...maybe even emotionally intelligent.

The edge isn't just a boundary – it's the frontier of human creativity. And once you lose it, no amount of AI can bring it back.

Do you really want 2025 on repeat, or are you ready to build something we haven't even imagined yet?

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.theaioptimist.com