Getty Images Unsuccessful in AI Infringement Claim

The long awaited judgment in the case of Getty Images v Stability AI was handed down yesterday in the Intellectual Property Enterprise Court. Unfortunately for Getty, the judgment was largely in Stability’s favour, with the exception of the judge finding that there was a “historic and extremely limited” trade mark infringement by Stability.

Getty had originally brought its claim against Stability for copyright infringement and trade mark infringement on the basis that it had allegedly used Getty’s images and text to train its generative AI model, and the outputs of that model resembled the original images and, on occasion, included Getty’s watermark.

During the course of the proceedings, Getty’s claim narrowed and, by the end of the trial, it had removed its copyright infringement claim in relation to the training of Stability’s AI model and the outputs (on the basis that the training was held to have taken place outside of the UK) as well as its claim for infringement of its database rights, thereby leaving Getty’s claims for trade mark infringement and the question of whether Stability’s AI model was itself an infringing copy (if it had been trained using copyright works) for the judge to decide upon.

On the remaining copyright point, the judge found that Stability’s AI model was to be considered an “article” within the meaning of the Copyright, Designs and Patents Act 1988. However, even though the model was found to be an article, Getty was unable to convince the judge that it was not an infringing copy.

On the trade mark points, the court had found that there was some evidence of real world users receiving outputs that contained Getty’s watermarks which could cause confusion, this was only substantive enough to prove infringement by early versions of the model within the meaning of section 10(2) of the Trade Marks Act 1994 (TMA). This was not substantive enough to prove infringement by later models within the meanings of sections 10(1), 10(2) or 10(3) of the TMA.

Getty’s main issue in its claim was that it had very limited UK based evidence to prove the alleged infringements and the fact that many of the outputs displayed infringing material in very early iterations of the model, which Stability later fixed.

The judgment is an astounding 219 pages long with some good guidance for those in trade mark practice, as well as those in creative and tech industries. The decision will be welcomed by those within the tech sector, given that AI has been coined “the next industrial revolution”. However, those within the creative sector will likely be disappointed given the recent discourse around AI developers using protected works to train their models without permission of the rightsholders. It will be interesting to observe whether Getty apply to appeal any elements of the judgment.

If you have any questions on the above, or if you need any advice or support in respect of your own intellectual property, please do not hesitate to contact the team at McDaniels Law on 0191 281 4000 or legal@mcdanielslaw.com.

Share the Post:

Related Posts