Sign in Subscribe
AI

Getty Withdraws Major Copyright Claims Against Stability AI as UK Legal Dispute Narrows

A screenshot representing Stability AI, the company at the center of a major copyright lawsuit with Getty Images.

Getty Images recently withdrew its primary copyright infringement allegations against Stability AI in the High Court of London, a significant shift in a high-profile legal case that has shaped the debate around copyright and artificial intelligence model training. While the case still proceeds on narrower grounds, this development underscores the complex, evolving landscape around content ownership in the era of generative AI.

The Background: Getty vs. Stability AI

In January 2023, Getty, a global leader in stock images, filed suit against Stability AI—the developer behind the AI-driven image engine Stable Diffusion—accusing the startup of utilizing millions of Getty’s copyrighted visuals to train its model without consent. Beyond the training practices, Getty claimed that the AI’s generated images sometimes closely resembled copyrighted content, even bearing visible Getty watermarks.

These core claims—the unauthorized use of images for training and direct copyright infringement in generated outputs—have now been dropped. Getty’s legal team cited insufficient evidence and a lack of expert testimony linking Stability’s actions directly to UK copyright law. Additionally, the court found no substantial overlap between AI-generated images and the originals.

Despite the withdrawal of the central copyright charges, Getty continues to pursue the case on secondary infringement and trademark grounds. The remaining arguments allege that the models themselves may act as infringing articles: if created outside the UK, but imported or used within the country, this could constitute a form of legal violation. Trademark claims relate to the alleged misuse of Getty’s branding or watermarks in AI outputs.

Stability AI, for its part, expressed satisfaction at the court’s decision, maintaining confidence that further allegations—such as those surrounding watermarks—will be unsuccessful, given that end users are unlikely to confuse such marks as genuine endorsements from Stability AI.

Getty’s legal efforts extend to the United States, where it seeks damages for alleged misuse of more than 12 million images by Stability AI, aiming for up to $1.7 billion in compensation. Meanwhile, the broader tech community is closely watching the outcome of a recent US court case in which a federal judge sided with Anthropic, another AI company, ruling that usage of copyrighted books in AI training did not automatically constitute copyright infringement.

Stability AI also faces further litigation, alongside companies like Midjourney and DeviantArt, in a case brought by visual artists over similar alleged copyright violations.

Getty’s Generative AI Strategy

Notably, Getty now operates its own generative AI solution that leverages a training set licensed from Getty’s vast stock photography and video archives. This tool enables users to create and license new images and creative content, demonstrating how legacy media companies are adapting to—and monetizing—AI-driven technologies.

Deep Founder Analysis

Why it matters

The Getty vs. Stability AI litigation marks a pivotal point for startups in the AI and content industries. It highlights the ambiguous boundaries of copyright law when applied to large-scale AI model training, making it highly relevant for founders building in generative AI, SaaS, and creative technology. With courts moving toward a narrowing of copyright claims, startups may find the legal risk landscape changing rapidly—impacting everything from dataset creation to go-to-market strategies.

Risks & opportunities

This case signals a dual-edged sword: risk for startups relying on unlicensed data, but opportunity for those who can successfully navigate (or broker) licensing deals for training sets. The Getty move also raises the bar for what constitutes substantive copyright harm in AI-generated content, perhaps lowering the threat of blanket litigation and encouraging innovation. A historical parallel can be seen in the rise of YouTube and music streaming, where rights management shifted toward content ID and monetization, not just litigation.

Startup idea or application

Inspired founders might explore creating a compliance and licensing platform for AI model training: a SaaS solution that matches, brokers, and monitors licensable data for generative models, ensuring provenance and legal cleanliness. Alternatively, startups can focus on AI-driven forensics to identify substantial similarity between AI outputs and copyrighted works—a tool valuable for both rights holders and AI developers.

AI Copyright Stability AI Legal Getty Images

Visit Deep Founder to learn how to start your own startup, validate your idea, and build it from scratch.

đź“š Read more articles in our Deep Founder blog.

Read next

Comments ()