Artists hit AI art generators with copyright lawsuit

AI image generators are facing their day in court.

Over the past year, tools like Midjourney and Stable Diffusion have stunned the internet by turning simple text prompts into polished artwork in seconds. But behind that magic lies an uncomfortable reality: these systems learned how to draw by ingesting billions of copyrighted images scraped from the web without permission.

Now, the creators whose work made that possible are fighting back with serious legal firepower.

The lawsuits hit the fan

Three artists—Sarah Andersen, Kelly McKernan, and Karla Ortiz—filed a class-action lawsuit against Stability AI, Midjourney, and art portfolio site DeviantArt.

They teamed up with lawyer and typographer Matthew Butterick alongside the Joseph Saveri Law Firm. If those names ring a bell, it is because they are the same legal team currently suing Microsoft, GitHub, and OpenAI over AI coding assistant Copilot.

The class action accuses the companies of direct and vicarious copyright infringement, DMCA violations, breach of DeviantArt terms of service, and violations of California's unfair competition laws. The core issue is that models like Stable Diffusion rely on datasets like LAION-5B, a 5-billion-image trove scraped largely without artist consent, which can then be prompted to mimic those same artists' distinctive styles.

Just days later, on January 17, 2023, commercial powerhouse Getty Images piled on. Getty took Stability AI to the High Court of Justice in London, alleging the company illegally scraped millions of copyrighted photos to train Stable Diffusion for commercial gain while bypassing established licensing options.

Fair use vs. free-riding

AI developers have long leaned on the defense of fair use under U.S. copyright law, arguing that training a neural network is transformative.

Butterick and the artists argue the opposite: training commercial models on copyrighted work without consent, credit, or compensation is flat-out exploitation. Their stated goal is to force the industry into establishing ethical guardrails.

Interestingly, Getty was careful not to look anti-technology. In its statement, the stock agency noted that AI has genuine potential to stimulate creative work and pointed out that it has already licensed its catalog to other AI companies. Getty's real issue is straightforward: Stability AI skipped the checkout line.

Why it matters

For artists and creators, these cases will define the boundaries of digital ownership in the AI era. If the plaintiffs win, scraping someone's portfolio to train a tool that undercuts their market could become legally unsustainable. If the AI firms prevail, creatives will have to accept that anything posted publicly is fair game for machine training.

For the AI industry, the era of treating the open web as an all-you-can-eat data buffet is likely coming to an end. Training frontier models may soon require massive licensing budgets, which ironically favors deep-pocketed tech giants over open-source upstarts like Stability AI.

We think the AI firms relied too heavily on a broad interpretation of fair use without preparing a backup plan. Getty's lawsuit in particular shows that legal licensing frameworks already exist; ignoring them to save money looks less like technological disruption and more like corporate corner-cutting that could cost them dearly.