This is the third in the series of
posts on the opening days of the Getty Images v Stability AI trial. The posts
covering the trade mark claims and copyright and database rights claims are here
and here.
Today we look at the defences raised by Stability.
Pastiche
It is worth mentioning to begin
with that Stability has dropped its pastiche defence. This is a shame in some
respects, as this would have added to the relative paucity of decisions on this
defence (like Shazam).
E-Commerce Safe Harbours
Stability relies on regulations 18
and 19 of the Electronic Commerce
(EC Directive) Regulations 2002,
which implemented articles 13 and 14 of Directive 2000/31/EC in
the UK, namely the caching and hosting defences. As the proceedings and
evidence have developed, Stability has limited its reliance on the hosting
defence to the ‘image2image’ case, i.e. the use of images by users of Stability,
with and without additional text prompts, to create new images, and Stability’s
role in that process. Stability seemed unclear as to whether there was any
allegation as to any caching, so proceeds on the basis that no discussion was
needed on caching.
![]() |
A neutral and passive cat |
Caching defence
This defence gets little attention
in Stability’s submissions on the basis that it considers Getty is not placing
reliance on any copying taking place in the course of transmission of the
outputs to users. It appears from Getty’s submissions that there is some
miscommunication on this point, but that if Stability is relying on the
defence, it is not entitled to. It focuses on the fact that the claim relates
to the provision of the outputs, not the transmission of them, the lack of “onward
transmission” of the alleged copies, that the storage is not automatic,
immediate and temporary, with the focus on the lack of transience of the storage,
and the modifications to the content undertaken by Stability. The scope for
this to form a significant part of the decision seems limited.
Hosting defence
Save in relation to where users
use Stable Diffusion on their own computer (i.e. not at distance), Stability
equates its position with that of Google, as set out in Google France,
i.e. that, like Google, it processes data entered by third parties (advertisers
and users in Google’s case, users in Stability’s case).
Taking into account what Stability
says is a limitation on Getty’s claim to the settings leading to the generation of images closely
similar to the inputted image, Stability plays no active role in creating the
new image. It compares its role to saving an image from one format into
another. Stability claims that those settings requiring very close imitation of
the input image are like “switching off the AI” such that the output image is
almost identical to the input image. Stability is shifting the responsibility
onto the user, of whose acts it has no specific knowledge, and who has uploaded
an infringing image.
Getty’s response is that Chapter
II, Section 4 of the E-Commerce Directive, which contains the caching and
hosting safe harbour defences, only applies to “intermediary service providers”,
that being a narrower subset of “information society services”. The defences,
Getty says, are limited to those who do nothing other than act as an information
society services (Papasavvas and Uber).
Getty says Stability is not an intermediary as it is a bipartite rather than
tripartite relationship; one cannot be an intermediary when there are only two
people. Getty also says that Stability goes well beyond a merely technical,
automatic and passive role. Getty relies on Google France, L’Oréal
and Swatch as showing that the point at which a platform goes from
technical, automatic and passive to not is quite low.
Onwards
So that brings to an end this
series of posts summarising main areas of dispute. There are other potentially
important points in this bountiful case, but they are best explored once we
have a judgment.
By way of update on the trial,
Getty appears* to have been given a chance by the Court of Appeal to appeal the
judge’s decision not to allow Getty to rely on ‘CSAM’ material. This involves a
partial adjournment to some of the evidence, and a bit of a disruption to the
trial timetable. We shall see what impact this has.
*Thanks again to Rebecca Newman for her helpful updates on LinkedIn
Getty Images v Stability AI - UK trial begins (Part 3 - Defences)
Reviewed by Oliver Fairhurst
on
Thursday, June 12, 2025
Rating:

No comments:
All comments must be moderated by a member of the IPKat team before they appear on the blog. Comments will not be allowed if the contravene the IPKat policy that readers' comments should not be obscene or defamatory; they should not consist of ad hominem attacks on members of the blog team or other comment-posters and they should make a constructive contribution to the discussion of the post on which they purport to comment.
It is also the IPKat policy that comments should not be made completely anonymously, and users should use a consistent name or pseudonym (which should not itself be defamatory or obscene, or that of another real person), either in the "identity" field, or at the beginning of the comment. Current practice is to, however, allow a limited number of comments that contravene this policy, provided that the comment has a high degree of relevance and the comment chain does not become too difficult to follow.
Learn more here: http://ipkitten.blogspot.com/p/want-to-complain.html