Getty drops part of AI claim against Stability

In Short: Getty Images has quietly dropped several key parts of its UK copyright lawsuit against Stability AI — including the headline-grabbing claim that 7.5 million of its images were used to train the Stable Diffusion model. What remains is a narrower dispute focused on whether 17 specific AI-generated images infringe Getty’s rights.

What’s Going On?
This is a major tactical retreat. Getty had originally claimed that Stability trained its AI using millions of its images scraped from the web — but it’s now dropped that core allegation. That means the court won’t be deciding the biggest legal question hanging over AI: is training a model on copyrighted content unlawful?

Instead, Getty will argue that 17 output images from Stable Diffusion are so similar to Getty’s that they amount to infringement. Stability, for its part, will argue that any similarities are incidental or the result of user prompts — not copying.

Insights: This move sharply limits the trial’s potential impact. What was shaping up to be a landmark case on the legality of AI training now looks more like a conventional copyright dispute over a small set of outputs. For AI developers, that’s a breather — no sweeping precedent (yet) on the legality of using scraped content for training. But the bigger copyright clash is far from over. With regulators circling and lawsuits stacking up elsewhere, the AI world is still waiting for the moment when a court draws the lines on what’s fair game for training machines.

How Does This Impact Law Firms?