The much-anticipated High Court proceedings between Getty Images and Stability AI is now underway with a selection of Getty's arguments being ruled inadmissible on the first day of trial. Getty immediately appealed that ruling to the Court of Appeal, which handed down its decision within minutes of the end of the parties' oral arguments last Monday (with the written decision made available very shortly afterwards).
The trial judge's ruling concerned the admissibility of allegations, appearing for the first time in Getty's written opening submissions dealing with "tarnishment" under s10(3) of the Trade Marks Act 1994, alleging that Stable Diffusion is capable of creating CSAM images (i.e. images depicting the abuse of children). Stability had been prepared to address examples of objectionable and offensive, but non-CSAM, images previously identified by Getty, but objected to the additional CSAM allegations being made so late in the proceedings and so close to trial, not least because of the difficulty of testing them in light of the statutory framework around the creation of such images.
The trial judge, Joanna Smith J, ruled Getty's CSAM allegations inadmissible on the basis that the material had not been pleaded within the portion of Getty's Particulars referring to "images that contain pornography, violent imagery, and propaganda".
The Court of Appeal swiftly rejected Getty's appeal of that ruling, commenting that while CSAM is, unsurprisingly, within the ordinary definition of "pornography", Getty had given no indication before serving its skeleton argument that it would be relying on CSAM at trial. The consequence was that Stability could not fairly be asked to respond to such material, and Getty cannot deploy it during the trial.
Getty was then ordered to pay Stability's costs of the appeal.
The sudden inclusion and rapid exclusion of CSAM material was an unexpected turn of events at the beginning of a trial whose oversimplified themes of "IP versus AI" are liable to overshadow the importance of procedure in litigation. The objection to the production of CSAM material bearing a watermark with even a passing resemblance to a party's trade mark is, of course, understandable. But so too is the objection to being accused of producing those images, particularly where the opportunities to interrogate that accusation are limited.
So, although these rulings will have given Stability a fillip, and imposed an additional financial burden on Getty, in reality the parties are simply put back to the positions they were in a few weeks ago, when pleadings closed.
The upshot, then, is that rather than fundamentally changing the direction of what will be a hugely important decision on the relationship between IP and certain types of AI models, this skirmish illustrates the equally important requirement of adhering to the court's procedural rules. These intricate and often dry provisions do not always make the headlines, but they are no less important than the claims whose conduct they control.
Anyone considering pursuing or defending allegations of IP infringement will be well-advised, therefore, to make sure they do not allow the big picture issues being pursued to obscure the minutiae of how the proceedings are to be conducted.
The trial is expected to continue for another week, with a judgment not expected until after the summer break.
To stay up-to-date with the latest policy and judicial activity in this area around the world, please visit our leading Generative AI – IP cases and policy tracker or reach out to any of Mishcon's Artificial Intelligence and Machine Learning Team.