Getty Images Copyright Lawsuit Kicks Off as UK Data Bill Suffers 5th Defeat

The case, which has produced 78,000 pages of evidence via discovery, could go a long way toward setting the parameters for the use of copyrighted works to train AI models before the two houses of Parliament are able to settle their differences over the Data Bill, U.K. legal experts say.

“Legally, we’re in uncharted territory. This case will be pivotal in setting the boundaries of the monopoly granted by UK copyright in the age of AI,” Rebecca Newman, a lawyer at Addleshaw Goddard, told Reuters. Cerys Wyn Davies, of Pinsent Masons, said the High Court’s ruling “could have a major bearing on market practice and the UK’s attractiveness as a jurisdiction for AI development.”

Getty accuses Stability of scraping its vast library of photos without permission to train the Stable Diffusion image generator. Stability counters that its tool is an engine of fair use and free expression and does not violate any of Getty’s rights.

In court filings, Stability said the lawsuit poses “an overt threat to Stability’s whole business and the wider generative AI industry.”

Getty disputes that framing, claiming in its own filings that the suit “is not a battle between creatives and technology, where a win for Getty Images means the end of AI. The two industries can exist in synergistic harmony because copyright works and database rights are critical to the advancement and success of AI … the problem is when AI companies such as Stability want to use those works without payment.”

Read more: Elon Musk Is Creating an OpenAI Competitor

The trial is expected to last several weeks. Getty has filed a parallel lawsuit against Stability in the U.S.

The trial is playing out against the backdrop of the government’s faltering effort to steer the Data (Use and Access) Bill through Parliament. The bill is key to Prime Minister Keir Starmer’s campaign to promote the U.K. as a leader center for technology innovation, including AI. But the sprawling measure has drawn ferocious pushback from the creative industries fearful that the bill, as written, would allow the use of copyrighted works without permission or payment to train AI models.

On Wednesday (June 4) the House of Lords voted 221-116 in favor of an amendment offered by Baroness Beeban Kidron to require AI companies to disclose all the data used to train a model. It was the fifth time the peers attempted to add a version of the amendment to the Data Bill. But the Labour-led majority in the House of Commons has rejected it each time at the urging of the government.

“It is not fair, not reasonable, not just, balanced or any other such word to stand in the way of the creative industries identifying those who are taking their work or their property,” Kidron said ahead of Wednesday’s vote. “We have asked privately and repeatedly on the floor of both houses what is the government going to do to stop the work of creatives from being stolen right now? The answer is nothing.”

If the peers and MPs cannot find a compromise, under Parliamentary convention, the bill would fall, a rare but not unprecedented development, The Guardian reported. Were that to happen, MPs in the House of Commons, which has primacy as the elected branch, could invoke the Parliament Act to pass the bill without the amendment in the next parliamentary session.

That would significantly delay the bill, however, by which time the High Court may effectively have decided the question of whether AI companies can use copyrighted works to train their models without first securing a license from the rights owner.