Embedded Tweets and Display Rights: Dangerous Legal Ground for the Web
Written by Stan Adams
In a troubling recent decision (Goldman v. Breitbart) a court in the Southern District of New York found that embedding an image from Twitter in a web page hosted by a news sites can infringe on the exclusive right of the photographer to control the public display of the image. In the case, photographer Justin Goldman said that new sites, including Breitbart, infringed on this right when they included an embedded image of a tweet that contained a photograph he took of Patriots quarterback Tom Brady in the Hamptons.
Putting aside the fact that the photographer could have used a more direct method of controlling the use of his image, like a DMCA 512 take-down notice, interpreting an embed (or even worse, a hyperlink) as a form of copyright infringement is a bad idea. The various forms of hyperlinking, including embedded content, form the connective tissue of the web. They provide instant connections between separate sites and various forms of content, allowing sites to offer a more streamlined experience and layered information. They are a crucial part of what makes the web open and accessible, which is why judges and policy makers should exercise caution when changing the legal landscape for links.
Disregarding the way technology works is a flawed approach to legal reasoning. For copyright and technology, the details matter.
To help understand why the recent ruling in Goldman is so troubling, here’s a brief explanation of how embedded links work. When you use a browser to access a website, the browser basically reads a set of instructions from the server hosting the website about where to find the bits and pieces of the site and how to assemble them. Some of those pieces will probably be stored on the same server as the website’s instructional code. Other pieces may be stored elsewhere, on different servers. The assembly instructions for websites with embedded content tell the browser where to look for the content (using a uniform resource locator or URL), and where to put it on the page. The browser then locates the proper server and asks it to send the file for the embedded content. If it agrees, the server sends the content directly to the browser on the user’s computer. The website’s server has no interaction with or control over the embedded content other than pointing to its address.
In 2007, the 9th Circuit considered whether Google infringed the display rights of photographers with its “framing” feature of Google Image Search. It reasoned that because Google did not host the images on its own server, but rather provided instructions (to browsers) to find the images on servers not under Google’s control, Google was not responsible for infringement. This became known as the “server test.” The legal reasoning applies the language of the Copyright Act to the technical details of how browsers interact with websites. The Court found that, since only the server hosting the linked content had control of the files, it alone was capable of “communicating” the image files and therefore Google’s embedded link did not constitute a “display” under the Act.
In the recent case, Judge Forrest declined to apply the “server test,” finding that “a website’s servers need not actually store a copy of the work in order to ‘display’ it.” But servers don’t “display” embedded files because they can’t “transmit” or “communicate” a file they don’t possess. There is an important technical difference between sending a copy of an image file and sending instructions for where to look for a file. In the latter case, the website sending “embed” instructions has no control over the third-party servers, or the content they host. In this case, the third-party servers were operated by Twitter, which, had they been asked, could have removed or blocked access to the offending tweets. Instead, they were publicly accessible to anyone with the proper URL, including Breitbart reporters.
There is a much smaller technical difference between embedding and linking, which is why the argument that “sending instructions is the same as sending a file” is so troubling. They both consist of instructions, one of which directs the browser to retrieve the content automatically (the embed) and the other one requiring a click (or even just a “hover”). In both cases, the actions are performed by the end user’s browser and the image file is “communicated” by the third party directly to the end user. Although Judge Forrest tries to distinguish hyperlinks as different based on the “volitional” element, it is unclear how variations on hyperlink technology might fit that reasoning. For instance, would a hover-to-show type link satisfy the volitional element? Even just the possibility of infringement liability for links could chill their usage, reducing the utility and the “depth” of the web.
The technical workings of the internet have, in many ways, developed in response to judicial interpretations of copyright law. Moving away from a technically detailed approach to the law’s application creates fundamental problems for the open web, and could discourage innovative new ways to create and share information. This is why technically detailed applications of law like the “server test” make sense; it is based on concrete and verifiable information and keeps the liability on the party with actual control over the copyrighted work. More generally, disregarding the way technology works is a flawed approach to legal reasoning. For copyright and technology, the details matter.