Redistributed

General

AI and the Creative Industries: A Redistribution Case Study

Published March 11, 2026

laborcreative-industriesartmusicwritingredistribution

If you want to understand how AI redistributes value, look at the creative industries. Not because artists are uniquely sympathetic — though they are — but because the mechanism is unusually visible. Human creators make work. AI companies scrape that work to train models. Those models generate competing output at near-zero marginal cost. The original creators lose income. The AI companies capture the surplus.

This is not speculative. It is happening now, measurably, across music, visual art, writing, film, photography, and design. Each industry tells a slightly different version of the same story, and together they constitute the clearest redistribution case study of the AI era.

The pattern also reveals something important: creative workers have mounted the most effective collective response to AI displacement of any sector so far. The tools of resistance — litigation, organizing, legislation, and technical countermeasures — are being developed here first, creating templates that workers in other industries can adapt.

Music: royalty dilution in real time

The music industry offers the most quantifiable evidence because streaming platforms produce detailed data on who gets paid and how much.

Suno, the leading generative AI music platform, produces roughly 7 million tracks per day. Those tracks enter the same streaming ecosystem as human-made music. Deezer reports that AI-generated tracks account for 39% of its daily uploads — approximately 60,000 per day — and considers 85% of streams on those tracks fraudulent. The implications for human artists are arithmetic: streaming services use pro-rata royalty models, where total payouts are divided by share of total streams. Every AI-generated track that accumulates plays draws from the same fixed pool. More AI content means each human stream is worth less.

The major labels responded with litigation. Universal Music Group and Sony Music Entertainment sued both Suno and Udio for copyright infringement, alleging that these platforms trained on copyrighted recordings without authorization. Warner Music Group took a different route, settling with Suno in November 2025 in exchange for licensing terms and revenue sharing. The artist community launched its own campaign: the "Say No to Suno" open letter, backed by the Music Artists Coalition, the European Composer and Songwriter Alliance, and hundreds of individual artists, called AI-generated music "brazen smash and grab."

The redistribution here is structural. Artists created the recordings that trained the model. The model now competes with those same artists for a finite royalty pool. The artists received nothing for the training and earn less from their own work as the pool dilutes. The broader question of Who Owns Training Data sits at the heart of every AI copyright dispute — and music shows what happens when it goes unresolved.

Visual art: the billion-image scrape

The visual art community was among the first to confront AI training at scale. Stable Diffusion was trained on LAION-5B, a dataset containing over five billion image-text pairs scraped from the public internet. Midjourney and DALL-E followed similar approaches, building commercial products on the collective output of millions of artists, illustrators, and photographers who never consented to the use.

The backlash was immediate. In late 2022 and early 2023, artists organized protests on ArtStation, flooding the platform with "No to AI Generated Images" banners. Getty Images sued Stability AI in both the United States and the United Kingdom, alleging that Stable Diffusion was trained on millions of Getty's copyrighted images without permission. A separate class-action suit, Andersen v. Stability AI, brought by artists Sarah Andersen, Kelly McKernan, and Karla Ortiz, is working its way through US courts and could set precedent on whether AI training on copyrighted work constitutes infringement.

The economic impact runs beyond lawsuits. Illustrators and concept artists report losing commission work to clients who now use Midjourney to generate images for a $10 monthly subscription. As Kate Crawford documented in Atlas of AI, the extraction of value from distributed human labor into concentrated corporate products is a recurring pattern in AI development — the visual art case simply makes the extraction visible because the outputs are recognizable.

Writing: the content flood

Generative AI hit writing markets with particular force because text generation was the first capability to reach commercial quality. ChatGPT's release in November 2022 triggered an immediate restructuring of content production.

The evidence is stark. A study in the Journal of Economic Behavior & Organization found that writing jobs on freelance platforms dropped 30% within eight months of ChatGPT's release. Content mills that once employed thousands of human writers pivoted to AI generation, using humans only for light editing. SEO content farms proliferated — able to produce thousands of articles per day at negligible cost, flooding search results with machine-generated text of varying quality.

Journalism was not spared. CNET, Sports Illustrated, Gannett, and others experimented with AI-generated articles, sometimes without disclosure. Layoffs across the media industry accelerated, with AI capabilities cited as a contributing factor alongside broader revenue pressures. The Writers Guild of America reported that its members' residual income had already been declining before AI, but generative tools created a new category of threat: not just fewer writing jobs, but the elimination of entire content categories that had sustained freelance careers.

The redistribution is direct. Human writers produced the text that trained the models. Those models now produce competing text. The value flows from individual creators to the companies operating the AI systems, and to the businesses that use those systems to replace human writers.

Film and video: Hollywood draws the line

The film and television industry produced the most consequential labor response to AI displacement to date. In 2023, both SAG-AFTRA (the actors' union) and the Writers Guild of America went on strike, and AI protections were central to both disputes.

The WGA won provisions ensuring that AI cannot be credited as a writer, studios cannot require writers to use AI tools, and AI-generated text cannot be used to undermine writers' credits or compensation. SAG-AFTRA secured consent requirements for digital replicas — studios must obtain explicit permission before using AI to replicate an actor's likeness, voice, or performance. These provisions applied to both existing performers and background actors, closing a loophole that would have allowed studios to scan extras once and use their likenesses indefinitely.

These were the first major labor agreements to establish protections against AI displacement. They matter as precedent: they demonstrated that organized workers can negotiate meaningful constraints on how AI is deployed in their industry. The details of how this was won — and what other sectors can learn from it — are explored in Collective Bargaining.

Meanwhile, the technology continues to advance. OpenAI's Sora, Runway's Gen-3, and Pika Labs' video generation tools are producing increasingly convincing short-form video. The immediate threat is to visual effects artists, animators, and stock footage creators — roles that are often freelance or contract-based and lack the union protections that SAG-AFTRA and WGA members negotiated.

Photography: platforms pivot, contributors lose

The stock photography industry illustrates how platforms can simultaneously profit from AI and undermine their own contributors. Getty Images and Shutterstock both launched AI image generation services — Getty partnered with Nvidia, Shutterstock with OpenAI — using their own contributors' images as training data.

The business logic is transparent. These companies built libraries over decades by paying photographers per image. They then used those libraries to train AI models that generate competing images on demand, at a fraction of the cost. Some contributors receive modest payouts from AI training funds, but the economics are clear: a photographer who once earned recurring royalties from a stock image now competes with an AI model trained partly on their own work.

This is the redistribution paradox that Ursula Franklin described in The Real World of Technology: when systems designed to serve a community of producers are restructured to serve the interests of the platform itself. The photographers' work was the raw material; the platform extracted and concentrated its value.

Design: the junior position disappears

In graphic design, the impact follows a familiar pattern of automation: routine tasks are absorbed first, eliminating the entry-level positions that traditionally trained the next generation of professionals.

Canva's AI features, Adobe Firefly, and a proliferation of AI design tools now automate tasks that once required a junior designer: generating layout variations, creating social media graphics, producing marketing collateral. Agencies report reducing junior headcount while expecting senior designers to incorporate AI tools into their workflows — producing more output with fewer people.

The long-term risk is a hollowing out of the talent pipeline. If junior design positions disappear, where do senior designers come from? The same question applies across creative fields: if entry-level writing, illustration, and production work is automated, the apprenticeship model that developed creative professionals for decades breaks down.

The redistribution mechanism

Across all six industries, the same four-part mechanism operates:

Training data extraction. Creators' existing work is used — typically without consent or compensation — to build commercial AI products. The training data is the raw material, and its creators receive nothing for it.

Output competition. The resulting AI systems generate content that competes directly with human-made work, often at near-zero marginal cost. The competitive advantage is not quality but price and speed.

Platform amplification. Distribution platforms — Spotify, YouTube, Instagram, stock photography sites — benefit from abundant cheap content regardless of its source. Their incentive is volume and engagement, not the welfare of human creators.

Value capture. The commercial value concentrates in AI tool companies (OpenAI, Stability AI, Suno, Midjourney) and in the businesses that use those tools to reduce labor costs. Original creators are cut out of the value chain entirely.

This pattern echoes what Daron Acemoglu and Simon Johnson describe in Power and Progress: technology does not automatically distribute its benefits broadly. Without deliberate institutional choices, the gains flow to those who control the technology while the costs fall on those whose labor it displaces. The Disruption Narrative — the idea that displacement is a necessary price of progress — deserves particular scrutiny here, because in the creative industries the "progress" is often marginal while the displacement is severe.

What comes next

The creative industries are testing every available response: litigation, legislation, collective action, and technical countermeasures.

On the technical front, researchers at the University of Chicago developed Glaze and Nightshade — tools that apply imperceptible perturbations to images, disrupting AI training while leaving the images visually unchanged. These are defensive measures, not solutions, but they give individual artists a degree of agency over how their work is used.

On the legal front, pending cases — Andersen v. Stability AI, UMG and Sony's suits against Suno — will clarify whether training on copyrighted material without permission constitutes infringement. The outcomes will shape the economics of every generative AI company.

On the policy front, proposals include compulsory licensing for training data (modeled on music's mechanical license), mandatory attribution for AI-generated content, opt-in frameworks requiring creator consent before training, collective licensing administered by creator organizations, and platform regulations requiring disclosure of AI-generated content.

The most ambitious proposals imagine structural alternatives: creator-owned AI Cooperatives, where artists collectively control models trained on their work and share in the revenue. Organizations like the Collective Intelligence Project are exploring governance models for shared AI resources. These remain largely theoretical at scale, but they represent the redistribution question in its purest form — not whether AI creates value, but who captures it.

It is worth acknowledging complexity here. Some individual creators are using AI tools to expand their own capabilities — musicians producing arrangements they couldn't afford to hire session players for, visual artists exploring styles beyond their training, writers prototyping ideas faster. The technology is not inherently extractive. The extraction happens when the value chain is structured so that creators' work trains models without compensation, and when platforms treat human and AI content as interchangeable commodities. The goal is not to reject AI in creative work but to ensure the terms are fair.

The creative industries are ground zero for this question because the value chain is visible and the stakeholders are organized. What happens here — the legal precedents, the labor agreements, the policy frameworks — will set the template for every other sector where AI competes with the humans it learned from.

Sources

  • Hui, Xiang, Reshef, Oren, and Zhou, Luofeng. "The Short-Term Effects of Generative Artificial Intelligence on Employment." Journal of Economic Behavior & Organization (2024). ScienceDirect
  • Vincent, James. "Getty Images sues AI art generator Stable Diffusion in the US for copyright infringement." The Verge, February 6, 2023. The Verge
  • Acemoglu, Daron and Johnson, Simon. Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity. PublicAffairs, 2023.
  • Crawford, Kate. Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press, 2021.
  • Shan, Shawn et al. "Glaze: Protecting Artists from Style Mimicry by Text-to-Image Models." University of Chicago, 2023. Glaze Project
  • "Say No to Suno: Open Letter." Music Technology Policy, February 23, 2026. Music Technology Policy
← Back to all articles
On this page