Posted by Peter Brown
In their early stages, computing and digitalization raised difficult and novel questions for courts interpreting copyright laws. Judges wrangled with copyright’s application to operating systems, graphical user interfaces, and other fundamental aspects of modern day computing. One core aspect of internet use is hyperlinking. Simply put, hyperlinks take you from one place on the web to another. Clicking on this hyperlink text will navigate you to another webpage (in this case, Wikipedia’s entry on hyperlinks). The interaction between hyperlinking and copyright law led to an interesting legal question: does linking to another website’s copyrighted content constitute infringement? This question was thought to be more or less settled in favor of finding non-infringement.
Commentators and practitioners alike have felt secure in the knowledge that hyperlinking and similar use of images (such as embedding) is not necessarily infringing. However, a recent New York district court decision, Goldman v. Breitbart, has turned certainty into doubt. If widely adopted, the court’s reasoning in Goldman v. Breitbart could seriously threaten the utility and structure of the internet.
Understanding the unexpected ruling in Goldman requires briefly exploring prior decisions relating to hyperlinks and embedding. In the notable case Kelly v. Arriba, the court considered a photographer’s copyright claim against a search engine for violating the photographer’s right to reproduce and display his photos. The search engine Arriba trawled the internet, downloaded images, created lower-resolution thumbnails, deleted the images it downloaded, and displayed search results in the form of the thumbnails. The court found Arriba’s use of thumbnails was for a completely different purpose than Kelly’s original photos and caused little to no harm to the market for Kelly’s originals. Therefore, the court held that Arriba’s display of thumbnails to users was non-infringing fair use of Kelly’s photos.
With the Arriba case and others, websites have generally felt comfortable providing hyperlinks to other websites. Even if those other websites hosted some amount of infringing content themselves, this would not usually pose a problem for the site providing hyperlinks. Barring extreme behavior such as promoting the infringing content, websites would likely have to know or have reason to know of infringements on the linked site to be liable.
In the context of this background, let’s consider the Goldman case. It features photographer Justin Goldman against various media companies, including Breitbart, Yahoo, the Boston Globe, and Gannett. The dispute centers around a Snapchat photograph Mr. Goldman took of Tom Brady. After Mr. Goldman added the photo to his Snapchat story and shared it (presumably privately rather than publicly), the image made its way onto Reddit and then Twitter. Soon after, the news organizations named in the case spotted tweets with the photo and embedded tweets containing the photo in their news stories.
Mr. Goldman, who claims he never publicly released or licensed his photograph, sued the news organizations for violating his exclusive display right in the image. In a surprising ruling, the judge found in Mr. Goldman’s favor. The judge’s ruling means the news organizations violated Mr. Goldman’s exclusive display right – despite never having obtained, created, saved, or otherwise possessed a copy of the photo. More surprising was the court’s statement that “the fact that the image was hosted on a server owned and operated by an unrelated third party (Twitter) does not shield them from this result.” (emphasis added).
The court distinguishes Mr. Goldman’s case from another dealing with internet images: Perfect 10, Inc. v. Amazon.com. In that case, the court addressed two issues: first, did Google image search results in thumbnail format violate the copyright holder’s display right? Second, did the full-size images appearing after a user clicked on a thumbnail infringe on the copyright holder’s display right? The court distinguished between the two types of search results based on where the image host was located. Google stored the thumbnail images on its own servers which the court found to be infringing. In contrast, the full-size images were displayed through in-line linking (similar to embedding) and were ruled not infringing because Google never stored them anywhere on its own servers. On appeal, the Ninth Circuit upheld the lower court’s ruling.
In doing so, the Ninth Circuit adopted the Server Test in which liability depends on the location of the images. A publisher hosting and providing access to another’s images stationed on the publisher’s own server would likely be found liable for copyright infringement. In contrast, a publisher embedding or linking to images hosted on a third-party server would not necessarily be directly liable.
The Goldman court declined to adopt the reasoning in Perfect 10 stating that (1) an infringer can display a work without possessing a copy of it and (2) the user’s role was important in Perfect 10 where Google displayed results based on the user’s input. In contrast, the Goldman court reasoned that photos in embedded tweets load without any user input. Of paramount importance to the Goldman court was the user’s lack of volition in causing the embedded photos to load. As a result, the Goldman court found the Server Test did not apply.
While the Goldman case is still in its relatively early stages, the potential ramifications of the decision are concerning. Broadly interpreted, the court’s holding stands for the idea that anyone on the internet embedding a photo can be liable for violating the copyright holder’s display right.
Consider the transaction costs in such a situation could incur. A news article may have several embedded images. For each image, the media company would have to somehow track down the copyright holder and negotiate a license. Applied across every article on the website, this requirement would become burdensome and expensive, let alone the potential for liability.
Plus, there is potentially great uncertainty about the original copyright holder’s identity in relation to a photo on the internet. In Mr. Goldman’s case, the news organizations would have had to ask Twitter, Reddit, and Snapchat users for their sources, obtained cooperative and truthful responses along the entire chain of questioning, and then negotiated a license agreement with Mr. Goldman after learning his identity. The sequence of events considered here assumes at least one of the Twitter, Reddit, or Snapchat users knew and could identify the copyright holder. In instances where the users do not know or do not cooperate, the news site would have to find the copyright holder without any direction. Each step reduces the chances any website would opt to embed images.
The Goldman court, concerned that Perfect 10 elides the distinction between the right to display a work the right to make copies of a work, itself elides another distinction: the difference between direct and contributory infringement. Direct infringement is when a person without authorization displays, reproduces, or distributes a copyrighted work. Contributory infringement is when an entity induces someone else’s infringement and substantially contributes to that infringement.
If anything, the stronger argument for liability here should lie in contributory infringement because the websites showing embedded images deliberately select which images to embed and induce website users to infringe directly by causing their browsers to display those embedded images. Admittedly, even the Goldman court suggests that such a use might qualify as fair because the photo here is used for news and commentary which are explicit statutory exceptions.
Another concerning portion of the Goldman opinion states that liability “should not hinge on invisible, technical processes.” The court ignores that embedded images display only when the user’s browser makes a request to the image host. Only the user and the site hosting the image (here, it’s Twitter) interact. The publisher (in this instance, news websites) does not handle the content displayed beyond facilitating the user’s access to the image. This cannot constitute a direct infringement of the display right by the news organizations when the entity requesting and displaying the image is the user.
The Ninth Circuit addressed this issue directly in Perfect 10 v. Amazon. That court’s reasoning on the issue of HTML instructions is helpful. By stating “[p]roviding HTML instructions is not equivalent to showing a copy” the Ninth Circuit demonstrates its understanding of the significance of the technical distinction between showing a copy and providing HTML instructions. Facilitating a user’s viewing of an image through HTML instructions “raises only contributory liability issues, and does not constitute direct infringement of the copyright owner’s display rights.” (emphasis added). This is precisely what the Goldman opinion missed.
The distinction between direct and contributory infringement has also been lost by at least one legal commentator, Scott Alan Burroughs. In scornfully arguing against the news organizations in the Goldman case, Mr. Burroughs dismisses the media groups’ positions as technobabble. He contends, as the Goldman court does, that the distinction between (1) hosting and displaying an image and (2) embedding an image is meaningless to users. Either way, the user’s browser loads the image for viewing.
Both the court and Mr. Burroughs are correct on one point: most users do not care a whit who hosts images or how the images are loaded. As insignificant as these facts may be to users, the Goldman court, and Mr. Burroughs, the law is concerned with technical distinctions between displaying a work, copying a work, and facilitating others’ infringements on a creator’s copyright. Mr. Goldman’s Snapchat friends are responsible for the unauthorized copying of his photo.
Suing one’s friends is uncomfortable so the news organizations are a more natural target for Mr. Goldman. As tempting a target they may be, this does not give courts a license to ignore the legal distinctions between display, copying, and direct and contributory infringement.
Mr. Burroughs and the Goldman court rightfully assert that artists should be appropriately paid for their work. At the same time, the legal system cannot ignore the dramatic changes the internet has wrought upon the world. Accessing, copying, and sharing information has never been so quick and so simple. Undergirding much of the linking, embedding, and sharing on the internet is the understanding that preventing others from accessing your content means opting out.
As various media and technology organizations have raised dire predictions about the Goldman decision and what it means for embedding images, sensible comments allaying these fears come from Mr. Goldman’s attorney himself. In discussing the case with Reuter’s reporters, Mr. Goldman’s attorney Kenneth Norwick noted what most have ignored or omitted: Goldman never wanted his photo to make it on to the wider internet. Mr. Norwick distinguishes between “embedding authorized tweets and those containing images never intended for publication.” These comments align with the opt-out understanding that search engines and websites have reached – the entire point of posting an image to Twitter is to have it shared, copied, and linked to. In contrast, a Snapchat communication has a different purpose.
Yet if Mr. Goldman’s lawyer is correct, the problem of determining an image’s source remains. The news organizations assumed the image they embedded was one that originated on Twitter and therefore could be freely shared and embedded. This is why a knowledge requirement is important: facilitating a user’s access to content should give rise to liability when the facilitator knew or had reason to know of the infringing nature of the content.