Artificial Intelligence Art Is Certainly Fair Use! Umm – Not So Fast

Stephen Carlisle March 1, 2023

As a general rule, I do not comment on pending litigation. Too much can happen on the way to a judgement or court ruling. Once there is a court ruling, which generates substantive law, IMHO that is the best time for commentary.

However, artificial intelligence art program creator Stability AI has been sued three times in recent months. Once in a class action in the United States, secondly in the UK by Getty Images, ¹ and a third lawsuit, also by Getty Images, filed in the US District Court in Delaware. ²

I enter into this fray in response to this sneering comment on the class action lawsuit by a spokesperson for Stability AI:

“Anyone that believes that this isn’t fair use does not understand the technology and misunderstands the law.” ³

Well then! Verily the gauntlet has been thrown! So let’s take a look, shall we?

Toto, We’re Not in Kansas Anymore, Part I

Copyright is a global institution, and the programs’ distribution is world-wide. The first of Getty’s lawsuits was filed in the United Kingdom. The UK, for example, does not recognize “fair use” as it exists in America. Instead, it recognizes a much more limited concept of “fair dealing.” And UK law has specific provisions regarding the sort of data mining engaged in by Stability AI, which has the effect of making SAI’s action infringing. Article 29(a) of the Copyright, Designs and Patents Act of1988 provides:

Copies for text and data analysis for non-commercial research

                “(1) The making of a copy of a work by a person who has lawful access to the work does not infringe copyright in the work provided that—

(a)the copy is made in order that a person who has lawful access to the work may carry out a computational analysis of anything recorded in the work for the sole purpose of research for a non-commercial purpose”

And

“(2) Where a copy of a work has been made under this section, copyright in the work is infringed if—

(a )the copy is transferred to any other person, except where the transfer is authorised by the copyright owner, or

(b )the copy is used for any purpose other than that mentioned in subsection (1)(a), except where the use is authorised by the copyright owner.” ⁴

So, the type of data mining engaged in by SAI seems to create massive liability on their part, as the purpose was clearly not “non-commercial.”

Toto, We’re Not in Kansas Anymore, Part II

Defenses of AI art as being “fair use” all start at the same point: the opinion of the Second District Court of Appeals in Authors Guild v. Google, Inc. ⁵ In that case:

“Through its Library Project and its Google Books project, acting without permission of rights holders, Google has made digital copies of tens of millions of books, including Plaintiffs’, that were submitted to it for that purpose by major libraries. Google has scanned the digital copies and established a publicly available search function. An Internet user can use this function to search without charge to determine whether the book contains a specified word or term and also see “snippets” of text containing the searched-for terms.” ⁶

There, the Court held this was fair use, ruling:

“Google’s making of a digital copy to provide a search function is a transformative use, which augments public knowledge by making available information about Plaintiffs’ books without providing the public with a substantial substitute for matter protected by the Plaintiffs’ copyright interests in the original works or derivatives of them.” ⁷

So far so good. Except that the Getty Images case was filed in the District Court in Delaware. Why Delaware? According to the complaint, Stability AI, Inc is incorporated in Delaware. And Delaware is in the jurisdiction of the Third Circuit Court of Appeals, not the Second Circuit Court of Appeals. ⁸ As persuasive as the Second Circuit’s reasoning and logic in the Google Books case might be, the Courts of the Third Circuit are under no obligation to follow the ruling or treat it as settled law. It can rule any way it wishes.

Toto, We’re Back In Kansas, But We’re Still In Trouble

Last but not least, even if the case were being heard in the Courts governed by the Second Circuit, the outcome of fair use is hardly the slam dunk that SAI thinks it is.

First, the opinion in the Google Books case itself admits that the case “tests the boundaries of fair use.” ⁹ Next, recall the Court’s ruling depended upon the finding that Google Books was not “providing the public with a substantial substitute for matter protected by the Plaintiffs’ copyright interests in the original works or derivatives of them.” ¹⁰

That is not the case here. Clearly the intent of the SAI program is to act as a market substitute for the works they have ingested in the training process. Since the SAI program cannot really “create” anything, it can only make use of what has been ingested into the database, it is the ingestion point which becomes infringing. If you’ve already got the entire Getty Images database for free in your SAI program, why go and get a license from Getty Images?

Or consider this, if the ingestion of millions of copyrighted images from Getty Images into the SAI program is indeed fair use, what is to prevent other image manipulation programs, say PhotoShop from offering the Getty Images database as well?

Next up is the failure to recognize that the Second Circuit made a significant retreat from its ruling in the Google Books case when it issued its opinion in Fox News Network, LLC v. TVEyes, Inc. ¹¹

“TVEyes has two basic modes. A ‘search’ function allows the subscriber to find video segments that contain certain buzzwords. The ‘watch’ function allows that same subscriber to view that same segment for up to 10 minutes of video, unaltered from its first performance by Fox News.” ¹²

Fair use, right? Nope.

“‘This [3^rd] factor clearly favors Fox because TVEyes makes available virtually the entirety of the Fox programming that TVEyes users want to see and hear. While ‘courts have rejected any categorical rule that a copying of the entirety cannot be a fair use,’ ‘a finding of fair use is [less] likely … when the copying is extensive, or encompasses the most important parts of the original.’ (citation omitted) In this respect, the TVEyes Watch function is radically dissimilar to the service at issue in Google Books.” ¹³

And this:

“Since the ability to re-distribute Fox’s content in the manner that TVEyes does is clearly of value to TVEyes, it (or a similar service) should be willing to pay Fox for the right to offer the content.

By providing Fox’s content to TVEyes clients without payment to Fox, TVEyes is in effect depriving Fox of licensing revenues from TVEyes or from similar entities. And Fox itself might wish to exploit the market for such a service rather than license it to others. TVEyes has thus ‘usurp[ed] a market that properly belongs to the copyright-holder.’ (citation omitted) It is of no moment that TVEyes allegedly approached Fox for a license but was rebuffed: the failure to strike a deal satisfactory to both parties does not give TVEyes the right to copy Fox’s copyrighted material without payment.” ¹⁴

According to the Getty Images complaint, SAI has copied more than 12 million images in the Getty images database, ¹⁵ which has copyright registration TXu002346096. ¹⁶ In addition, for the purposes of trial, 7,216 individual images will be at issue. ¹⁷ It is easy to see how the inclusion of the entire Getty Images database into the training of the program will negatively affect the ability of Getty Images to license these images. Indeed, the complaint shows instances of SAI output with the Getty Images watermark still attached to them. ¹⁸

The issue is not what the output might look like, even though SAI and its apologists try very hard to make this the crux of the case. The issue is that an entire database was copied, and made available to the public without compensation to the copyright owner, just like the TV Eyes case.