The recent High Court ruling in favor of Stability AI in Getty Images’ copyright infringement case was “hardly surprising,” said technology lawyer Matthew Holman.
This week, London-based generative AI group Stability won a landmark case over the legality of using copyrighted data in training processes.
While seen as a blow to many copyright holders who fear AI companies will be able to use whatever IP they like to train models, the decision was not a shock to Holman, a partner at law firm Cripps.
Holman noted that the UK government has expressed its intention to “allow AI developers to train with their own material unless rights holders opt out” and, although it has not yet legislated this, it is currently engaged in consultation on the issue.
While the case can be seen as a landmark decision, Holman points out that it is “just the beginning of a complex battle between AI model developers and copyright holders.”
“The international landscape remains similarly complex, and the landmark case of New York Times v. Open AI, with underlying facts very similar to those of Getty Images, is likely to be decided next year,” he said.
Ed Newton Rex, a former senior employee at Stability AI and now chief executive of Fairly AI at Fairly Trained, a nonprofit copyright practice, called the court’s decision “very disappointing.”
“The law is clear: if you import it into the UK, that copy is an infringing copy and ‘if it was manufactured in the UK you would have infringed the copyright in the work in question’. That is clearly the case here,” he said.
“The scope of this case is clearly limited as the primary infringement claim was dismissed because Getty could not prove that the development took place in the UK. However, it is still disappointing and we hope it will be appealed.”
read more: Most generative AI models are likely to be ‘illegal,’ says former stability vice president
