A Deep-Research Guidebook for Visual Creatives

The Creative's
Honest Guide to AI

How to protect your work, choose the right tools, and decide — on your own terms — whether AI has a place in your practice.

Written for the artist who's been lying awake at 2am wondering whether the thing they love is about to be automated out of existence — and who deserves a straight answer, not a sales pitch.

Chapter 1

The Heist Nobody Asked Permission For

Let's start with what actually happened, because a lot of the noise around AI art has buried the real story under layers of either panic or dismissal.

In 2022 and 2023, a wave of AI image generators hit the market — Midjourney, Stable Diffusion, DALL-E 2, and others. They were genuinely astonishing. You could type "a melancholy fox in a rainstorm, Studio Ghibli style" and get something that looked like it took a skilled artist three days to paint. The tech community lost its mind with excitement. The creative community lost its mind with something else entirely.

Here's why: those models were trained on billions of images scraped from the internet. Without permission. Without compensation. Without so much as a notification to the artists whose work was included. The dataset most commonly used — LAION-5B — contained approximately 5.85 billion image-text pairs, and researchers who dug into it found the work of thousands of living artists, including some of the most distinctive and recognizable names in digital illustration.

What made it personal wasn't just the scale. It was the specificity. Artists discovered that you could type their name into Midjourney and get a convincing imitation of their style. Kelly McKernan, a fine artist known for her ethereal, otherworldly paintings, found that her name had become one of the most-used style prompts on the platform. Karla Ortiz, a concept artist who has worked on major Hollywood productions, found her work in the training data without her knowledge. Sarah Andersen, the cartoonist behind "Sarah's Scribbles," found that her distinctive style — built over years of personal creative development — was being replicated on demand.

In January 2023, the three of them filed a class action lawsuit against Stability AI, Midjourney, and DeviantArt. It was the opening shot of what has become a long legal war.

Where the Lawsuits Stand Right Now

The legal landscape as of early 2026 is genuinely unsettled, and anyone who tells you otherwise is either misinformed or selling something. Here is what we actually know.

Andersen v. Stability AI is the case to watch. After the court largely dismissed the initial complaint in October 2023, the artists refiled. In August 2024, a California judge ruled that the direct copyright infringement claims could proceed — a meaningful partial victory that kept the case alive. The trial is now set to begin on September 8, 2026. The discovery phase is underway, which means both sides are now compelled to share evidence. What comes out of that process could reshape the entire industry.

Getty Images v. Stability AI produced a landmark ruling in the UK on November 4, 2025. The High Court rejected Getty's primary copyright claim — finding that the Stable Diffusion model itself is not an "infringing copy" under UK law. But the court upheld a trademark infringement claim, finding that Stability AI's outputs had reproduced Getty's watermarks.

As of October 2025, there were 51 active copyright lawsuits against AI companies in the United States. Three were on appeal. Two had settled. The legal community's consensus: no definitive fair use ruling is expected before 2026 at the earliest.

What this means for you: the law is still being written. You are living through the period when the rules are being established. That's uncomfortable — but it also means your voice, your advocacy, and your choices as a consumer of AI tools actually matter right now in a way they won't once the precedents are set.

The Human Cost Nobody's Talking About

The lawsuits are important. But the human cost of what happened is something the legal filings don't fully capture. The Concept Art Association surveyed its members and found that the entertainment industry had already begun replacing entry-level concept artists with AI-generated imagery. The jobs that used to serve as the first rung on the ladder for young artists breaking into film and game production were disappearing. Not because the artists weren't good enough. Because a client could get "good enough" for free.

The ArtStation protest of December 2022 — when thousands of artists flooded the platform with "No AI Art" posts — was not a tantrum. It was a community in genuine distress, watching the economic floor of their profession drop out from under them. That distress is real. It deserves to be named, not dismissed. And it is the starting point for any honest conversation about what to do next.

Chapter 2

Not All AI Is the Same

Here is something the loudest voices on both sides of this debate tend to skip over: the AI tools available to you are not all built the same way, trained on the same data, or governed by the same policies. Treating them as a monolith is like saying "cameras are bad for photographers" because some photographers use hidden cameras to invade people's privacy.

The question is not whether to use AI. The question is which AI, how, and under what conditions.

The "Walled Garden" Tools — Built for Commercial Safety

Adobe Firefly is the most commercially defensible AI image tool currently available. Adobe has been explicit about its training data: Firefly was trained on 375 million+ high-resolution Adobe Stock assets and public domain content — content Adobe had legal permission to use. More importantly, Adobe provides indemnification to business users, meaning that if a client is sued over Firefly-generated content, Adobe covers the legal costs. For commercial work, this is a significant protection.

The honest caveat: Adobe Stock contributors cannot opt out of having their images used for AI training. This has generated real backlash — a "strike Adobe" movement emerged in October 2024. Firefly is commercially safe for the person using it. It was built, in part, on the work of people who had no choice. That tension is worth knowing about.

Procreate has taken the most unambiguous stance of any major creative tool: they have publicly committed to never incorporating generative AI features into their platform. For artists who want a digital drawing environment completely free of AI entanglement, Procreate is the answer.

The "Open Frontier" Tools — Powerful, Complex, and Politically Charged

Stable Diffusion (developed by Stability AI) is the open-source model at the center of most of the legal controversy. Its training data is the most problematic from an artist-rights perspective. However, because it is open-source, it can be run locally on your own computer — meaning your work never leaves your machine. Artists like Minta Carlson have used this to their advantage by fine-tuning Stable Diffusion on their own work, creating a private AI assistant that knows their style and generates content consistent with it — without feeding anything back to the company.

Midjourney remains the most aesthetically sophisticated of the major text-to-image tools. Its outputs are genuinely beautiful. Its training data practices are among the most contested. For commercial client work, the legal exposure is real. For personal creative exploration, the risk calculus is different.

Runway ML is the leading AI video generation platform. Their data security page emphasizes that user-generated content is not used to train their models by default. For video work specifically, Runway has become the tool of choice for artists who want professional-grade AI video without the ethical baggage of the image-generation controversy.

The Ethical Certification to Look For

Fairly Trained (fairlytrained.org) is an independent organization that certifies AI models trained exclusively on content where creators gave explicit consent. When you see a Fairly Trained certification, it means the model was built without scraping. This is the gold standard. Check for it when evaluating any new tool.

Chapter 3

The Artists Who Made the Jump — And What They Actually Said About It

This is the chapter that matters most. Not the legal analysis, not the tool comparison chart. The human stories. Because the most persuasive argument for anything is watching someone you respect do it and tell you honestly what it was like.

These are real people. These are their real words.

He is not describing a shortcut. He is describing a fundamental shift in how the creative process works — one where the gap between imagination and execution has collapsed. The vision comes first. The tool serves it. When critics called his Vogue cover "disgusting" because he created it with prompts, Barbera's response was measured but firm: "You can embrace it or keep just hating it, but eventually you'll lose your job or get replaced by someone who knows how to use it." That's not a threat. That's a professional reading of where the market is going.

AI closed that gap. Using Runway ML and Stable Diffusion's Deforum animation tool, he uploads footage he records himself, then animates it using prompts like "traditional Japanese watercolor style." The result is a visual language he could never have achieved alone — not because he lacked the vision, but because he lacked the technical execution skills. Ideas that used to collect "digital dust" in his notebook are now getting made.

And the outcome that surprises people most: she has increased her painting output as a result. The AI didn't replace her painting. It fed it.

She still has a full-time job. She creates after work hours. And every day she gets messages from people telling her that her video made them "feel" something. The discipline behind the output is entirely human. The emotion it produces in viewers is entirely real.

The Skeptic's Honest Reckoning

Not every artist who has engaged seriously with AI has arrived at enthusiasm. And that is worth honoring too. In June 2025, the New York Times Magazine published a personal essay by illustrator Christoph Niemann who described his full emotional journey with AI. It is one of the most honest pieces of writing about this moment that exists.

The Visionaries: Refik Anadol, Holly Herndon, and the Artists Redefining What's Possible

Beyond the working creatives navigating AI in their day-to-day practice, there are artists who have made AI the central medium of their work — and whose achievements have forced the art world to take the question of AI authorship seriously.

Refik Anadol creates immersive installations that transform massive datasets — weather data, architectural blueprints, brainwave activity — into emotionally resonant visual environments. His work has been exhibited at MoMA, the Serpentine Gallery, and venues around the world.

Holly Herndon composes music with an AI entity she named "Spawn," trained on samples of her own voice and others. She collects her own data, trains her own model, and co-creates with the machine at every step. She is also one of the most vocal advocates for artist consent in AI development — demonstrating that you can embrace the technology while fighting for the rights of the humans whose work feeds it.

Sougwen Chung, a former MIT research fellow, paints alongside robotic systems she built and programmed herself — called D.O.U.G. (Drawing Operations Unit Generation). Her work is not AI-assisted in the passive sense. She is in active dialogue with the machine. These artists are not cautionary tales. They are proof of concept.

New Voices

Animators & Game Artists on AI

Animation and game art are two of the most labor-intensive creative fields on earth. A single frame of hand-drawn animation can take hours. A single game environment can require months of concept iteration before a single polygon gets placed. So when AI tools arrived promising to compress that timeline, the reaction in these communities was — predictably — complicated. Here are the people actually living through it.

What makes Blaise's position so valuable is that it refuses the binary. He is not saying "AI is fine, stop worrying." He is not saying "AI will destroy us all." He is saying: the tool matters less than how it's used, who controls it, and whether the humans whose work feeds it are treated with dignity. That's a position a working creative can actually build on.

By 2023, he was using Runway Gen3 and Pika for AI video backgrounds on a Confluent conference keynote opener — replacing stock footage with generative video for the first time on a client project. His current toolkit spans Midjourney, Runway, ChatGPT, Veo, Seedance, Wan, After Effects, Premiere, and ElevenLabs. He is not using AI to replace his team. He is using it to do things that simply weren't possible before — like designing content for a 103-meter ceiling in an afternoon.

His advice to other motion designers: stop waiting for the technology to feel "ready." It will never feel ready. The artists who are building fluency with these tools right now are the ones who will be positioned to lead when the industry fully shifts. And it is shifting.

Her advice for artists starting out with AI: watch how other artists work first. Understand what you actually admire about their results. Then build your own workflows from the ground up — start simple (text-to-image), then add complexity (image-to-image, masks, ControlNet). The foundation matters. And remember: the goal is not to let AI make the art. The goal is to let AI handle the parts that drain you, so you have more energy for the parts that only you can do.

The Honest Dissent: When AI Makes the Job Harder

Not every game artist's experience with AI has been one of expanded possibility. Some of the most important voices in this conversation are the ones describing how AI is being imposed on them — not adopted by them. That distinction matters enormously.

Canavan and Hu are not anti-technology. They are pro-craft. And their objections point to something important: the question of who controls AI adoption matters as much as the technology itself. When artists choose to use AI tools on their own terms — to expand what they can make, to handle the tedious parts, to explore faster — that's one thing. When AI is imposed on them by clients or management to constrain their creative contribution and reduce costs, that's something else entirely. The difference is agency. And agency is exactly what this guide is trying to help you protect.

Chapter 4

The Tools That Fight Back

Understanding the threat is one thing. Doing something about it is another. Here are the concrete tools and strategies available to you right now.

Glaze: The Cloak

Glaze was developed by Ben Zhao and the SAND Lab at the University of Chicago. It launched in March 2023 and has been downloaded more than 6 million times. How it works: Glaze adds "barely perceptible perturbations" to your image's pixels — changes so small the human eye cannot detect them, but which cause AI models to misread your style. If an AI tries to learn your style from a Glazed image, it learns something wrong. The cloak protects the style signature that makes your work yours.

The origin story of Glaze is worth knowing. Zhao was invited to a Zoom call in November 2022 hosted by the Concept Art Association. Artists shared how they were being hurt by the AI boom. Karla Ortiz said: "I would love a tool that if someone wrote my name and made a prompt, like, garbage came out. Just, like, bananas or some weird stuff." That was the moment Zhao decided to build it.

Glaze is free. Download it at glaze.cs.uchicago.edu. Shawn Shan, the lead developer, was named MIT Technology Review's Innovator of the Year for 2024.

Nightshade: The Poison

Nightshade is Glaze's more aggressive sibling. Where Glaze is defensive — protecting your style from being learned — Nightshade is offensive. It adds an invisible layer of "poison" to your images that can actively damage AI models that scrape them. The numbers are striking: with just 50 to 200 poisoned images, Nightshade can visibly distort a trained AI model. It has been downloaded more than 2 million times.

His goal is not just to protect individual artists — it's to "slowly tilt the balance of power from large corporations back to individual creators." Nightshade is free. Download it at nightshade.cs.uchicago.edu.

Content Credentials: Your Digital Signature

The C2PA (Coalition for Content Provenance and Authenticity) standard is an open technical framework for embedding verifiable metadata in digital files. Developed by Adobe, Microsoft, BBC, Intel, and others, it allows you to attach a cryptographically signed record to your work that travels with the file — documenting who created it, when, with what tools, and any edits made.

Adobe Photoshop and Lightroom now support embedding Content Credentials natively. Several camera manufacturers — including Leica, Sony, and Nikon — are beginning to embed C2PA credentials at the point of capture. You can verify any image's credentials at contentcredentials.org.

Contract Riders: The Legal Fence

The Concept Art Association has developed model contract language that explicitly prohibits clients from using your work for AI training. If you do commercial work, add this language to your contracts now. Not because every client will comply. But because it establishes your position, creates a paper trail, and signals to the market that you take this seriously.

Copyright Registration: The Foundation

In the United States, copyright registration is required before you can sue for statutory damages — which can be up to $150,000 per infringement for willful violations. The US Copyright Office registration fee is $65 for a single work, $45 for a group of unpublished works. The time investment is minimal. The protection is significant. Go to copyright.gov.

Chapter 5

What the Schools Are Saying

If you want to understand where the professional creative world is headed, watch what the art schools are doing. Because they are the institutions that have to make a bet on the future — and right now, they are betting on AI literacy.

The School of Visual Arts (SVA) in New York City has issued an official statement that is worth reading carefully. Their position: "We recognize that generative AI literacy must become an institutional learning outcome. All students should be graduating with a working knowledge of the benefits and hazards of generative AI, particularly the ethical implications."

They frame it with a quote from conceptual artist Sol LeWitt that cuts to the heart of the matter: "Some artists confuse new materials with new ideas." That is the challenge. AI is a new material. The ideas still have to come from you.

ArtCenter College of Design in Pasadena allows faculty to use generative AI "when appropriate in support of their teaching and instruction" while requiring the same standards of citation as for any other source. The Cleveland Institute of Art has positioned itself as a leader in AI exploration in art and design education, actively integrating AI tools into the curriculum while maintaining the foundational emphasis on traditional craft.

The signal from the institutions is consistent: AI literacy is now a professional requirement. The question is not whether you will encounter AI in your career. The question is whether you will encounter it on your own terms or someone else's.

Chapter 6

The Numbers That Should Get Your Attention

The uncomfortable truth embedded in all of this data: the artists who are thriving in this environment are not the ones who refused to engage. They are the ones who engaged on their own terms, with their own creative foundation intact, and who used AI to amplify what they already knew how to do.

Chapter 7

The Historical Argument — Why This Has Happened Before

Every generation of artists has faced a technology that was supposed to make them obsolete. Every single one. When photography was invented in 1839, painters were told their profession was finished. What actually happened: photography pushed painters toward modernism. Impressionism, Expressionism, Cubism, Abstract Expressionism — all of these movements were, in part, responses to the question "if a camera can capture reality, what is painting for?" The answer turned out to be: everything photography can't do.

When digital photography arrived, film photographers were told their craft was dying. What actually happened: film photography became more valuable, not less. The scarcity and the intentionality of shooting on film became a feature, not a bug. When Photoshop arrived, illustrators were told their hand-drawn work was obsolete. What actually happened: the best illustrators found that their hand-drawn work became more distinctive and more valuable in a world flooded with digital imagery.

That is the argument. Not that AI is good or bad. But that the artists who have always survived technological disruption are the ones who understood their own value deeply enough to find what the new technology couldn't do — and then did that, brilliantly.

Chapter 8

How to Figure Out Where You Stand

Nobody gets to make this decision for you. Not this guide. Not the artists profiled in it. Not the art schools, the lawyers, the tech companies, or the critics. But here is a framework that might help.

Question 1: What is the irreducible human element in your work?

Before you decide anything about AI, get clear on this. What is it that you do that comes from your specific life, your specific vision, your specific way of seeing? Not your technical skills — those can be replicated. Your perspective. Your obsessions. Your voice. The irreducible human element in art is the intention behind it — the fact that a specific person, with a specific life and a specific set of experiences, chose to make this thing. AI cannot have that. It can simulate it. It cannot have it.

Question 2: What would AI actually do for your practice?

Be specific. Not "it would make things easier" — that's too vague. What specific bottleneck in your creative process would AI address? For Matty Shimura, it was animation. For Natalya Shelburne, it was time. For Hugo Barbera, it was iteration speed. For Minta Carlson, it was scale. If you can name the specific bottleneck, you can evaluate whether AI actually addresses it — and whether the tradeoffs are worth it.

Question 3: Which tools align with your values?

This is not a trivial question. The tool you choose is a statement about what you believe. Using Adobe Firefly says something different from using Stable Diffusion. Using a locally-run model trained on your own work says something different from both. You don't have to make a perfect choice. But you should make a considered one.

Question 4: What protections are you putting in place regardless?

Whether you decide to use AI or not, the following are worth doing: Download Glaze and apply it to work you post publicly. Register your most important works with the US Copyright Office. Add AI training prohibition language to your commercial contracts. Embed Content Credentials in your digital files. Stay informed about the Andersen v. Stability AI trial. These are not positions on AI. They are professional hygiene in a world where your work is being scraped whether you like it or not.

A Final Word

The artists in this guide are not naive. They know what happened. They know that their work, and the work of thousands of artists like them, was used without consent to build systems that now compete with them commercially. They haven't forgotten that. But they also know that the technology exists, that it is not going away, and that the choice is not between a world with AI and a world without it. The choice is between engaging with it on your own terms or having it happen to you.

All three of them are right. And all three of them are talking about the same thing: the irreducible, inefficient, gloriously human act of making something that matters because you made it. That is not going away. It cannot be automated. And it is, ultimately, the only thing that has ever made art worth having.

Appendix A

Your Action Checklist

Regardless of where you land on AI, these steps protect you and your work right now.

Appendix B

Resources at a Glance

Appendix C

Source Citations

1. 1.Observer, "Despite Creative Hubris, Artists Are Quietly Embracing A.I. in Their Work," Aaron Mok, December 5, 2024
2. 2.TIME Magazine, "Why This Artist Isn't Afraid of AI's Role in the Future of Art," Lily Marks, May 5, 2025
3. 3.Pro Video Coalition, "AI Tools Artist Spotlight: an Interview with Kelly Boesch," Jeff Foster, January 5, 2026
4. 4.DIY Photography, "This fashion photographer shot magazine covers, then quit her job to become an AI creator," Udi Tirosh, March 17, 2024
5. 5.Observer, "Artificial Intelligence as Co-Creator: Rethinking Art and Authorship," Dr. Alfredo Cramerotti and Auronda Scalera, May 12, 2025
6. 6.New York Times Magazine, "Sketched Out: An Illustrator Confronts His Fears About A.I. Art," Christoph Niemann, June 23, 2025
7. 7.MIT Technology Review, "The AI lab waging a guerrilla war over exploitative AI," Melissa Heikkilä, November 13, 2024
8. 8.PwC 2025 Global AI Jobs Barometer — "AI linked to a fourfold increase in productivity growth and 56% wage premium," June 3, 2025
9. 9.Upwork Research: "Upwork Research Reveals New Insights Into the AI-Human Work Dynamic," July 9, 2025
10. 10.Lightcast Report, "Beyond the Buzz: Developing the AI Skills Employers Actually Need," July 24, 2025 — analysis of 1.3 billion job postings
11. 11.NYU JIPEL, "Andersen v. Stability AI: The Landmark Case Unpacking the Copyright Risks of AI Image Generators," December 2, 2024
12. 12.Latham & Watkins, "Getty Images v Stability AI: English High Court Rejects Secondary Copyright Claim," November 13, 2025
13. 13.Getty Images newsroom, "Getty Images issues statement on ruling in Stability AI UK litigation," November 4, 2025
14. 14.Adobe newsroom, "Adobe Expands Generative AI Offerings Delivering New Firefly App," February 12, 2025
15. 15.School of Visual Arts, "Artificial Intelligence (AI) Statement and Guidelines," sva.edu
16. 16.ArtCenter College of Design, "ArtCenter Position and Policy on Generative AI," artcenter.edu
17. 17.Staffing Industry Analysts, "Generative AI boosts freelancers' pay, new survey finds"
18. 18.Glaze project page, University of Chicago SAND Lab — glaze.cs.uchicago.edu
19. 19.Nightshade project page, University of Chicago SAND Lab — nightshade.cs.uchicago.edu
20. 20.Content Authenticity Initiative, "How it works" — contentauthenticity.org
21. 21.Baker Hostetler Case Tracker: Artificial Intelligence, Copyrights and Class Actions — bakerlaw.com

Living Document

This guide grows with you.

The artists profiled here were willing to tell the truth about their experience. Now it's your turn. Whether you've embraced AI, pushed back against it, or landed somewhere complicated in the middle — your story belongs here. Every submission is reviewed and published as-written.