Stability AI wins UK copyright case against Getty Images

Stability AI has won a copyright case against Getty Images after a high court judge ruled that the firm did not break UK copyright rules when it trained its model on photographs taken from the internet.

It was a case that many artists were watching because it was one of the first proper tests of whether companies building machine-generated images could use already published work without first asking permission.

Getty Images had accused Stability AI of helping itself to the work of others in bulk, and said many of its pictures were taken to teach the model how to respond to text prompts.

Getty said Stability had acted as if artists’ work was a kind of free quarry and that the firm was careless about what was poured into its systems.

However, Getty dropped part of its case halfway through the summer trial because there was no proof that the training took place on UK soil.

Stability AI’s victory shows how hard it is for copyright owners to win these cases

Even so, the ruling was not a clear win for tech, as the judge did find trademark infringement in certain cases where Getty watermarks were visible in the images that Stability’s system made. The ruling also did not accept the bigger view that a trained AI model itself is an illegal copy.

The high court judge wrote that if a system does not literally store or reproduce the pictures it trained on, then it cannot be called an infringing copy under UK law. She stopped short of saying whether Stability was guilty of passing off and left that point on the table.

Stability’s legal success was significant because it showed just how hard it is for copyright owners to win these cases if training happens somewhere outside the UK or if the very idea of what counts as a copy has changed in this new digital era. Stability’s directors include James Cameron, the filmmaker who was behind Avatar.

The ruling will have brought relief inside the company. Its general counsel said the company was pleased with the verdict and claimed the core of the copyright case was now closed.

Getty was less content, and it said it remained deeply troubled that even a company of its size and means had found it so difficult to fight for the rights of creators due to the way the law is currently written.

It argued again that creators need more transparency so they can know which companies are using their work, and how. It is calling on governments to write stronger rules so that companies can avoid years of costly court battles.

The decision also throws more fuel onto a political row that Labour ministers have been trying to navigate. Artists like Elton John, Dua Lipa, Kate Bush, and the author Kazuo Ishiguro have all been lobbying for firmer copyright protections.

They fear a world where huge language models and image models take the market for creative work and leave the original artists with little income. Tech firms, on the other hand, say they will not be able to make systems that match the best in the world if they can only train on small pools of licensed content.

The government admits that uncertainty is now hurting both sides of the economy, Britain’s creative industries and Britain’s tech ambition, and says the current legal fog cannot go on.

One idea on the table is a change to the law that would allow copyrighted material to be mined for AI training automatically unless the rights holder makes an active choice to block such use.

Several lawyers who specialise in this area say this would be a major shift and would give the UK a far more open model than many countries in Europe.

Claim your free seat in an exclusive crypto trading community - limited to 1,000 members.