Rejecting years of settled precedent, a federal court in New York has ruled [PDF] that you could infringe copyright simply by embedding a tweet in a web page. Even worse, the logic of the ruling applies to all in-line linking, not just embedding tweets. If adopted by other courts, this legally and technically misguided decision would threaten millions of ordinary Internet users with infringement liability.

This case began when Justin Goldman accused online publications, including Breitbart, Time, Yahoo, Vox Media, and the Boston Globe, of copyright infringement for publishing articles that linked to a photo of NFL star Tom Brady. Goldman took the photo, someone else tweeted it, and the news organizations embedded a link to the tweet in their coverage (the photo was newsworthy because it showed Brady in the Hamptons while the Celtics were trying to recruit Kevin Durant). Goldman said those stories infringe his copyright.

Courts have long held that copyright liability rests with the entity that hosts the infringing content—not someone who simply links to it. The linker generally has no idea that it’s infringing, and isn’t ultimately in control of what content the server will provide when a browser contacts it. This “server test,” originally from a 2007 Ninth Circuit case called Perfect 10 v. Amazon, provides a clear and easy-to-administer rule. It has been a foundation of the modern Internet.

Judge Katherine Forrest rejected the Ninth Circuit’s server test, based in part on a surprising approach to the process of embedding. The opinion describes the simple process of embedding a tweet or image—something done every day by millions of ordinary Internet users—as if it were a highly technical process done by “coders.” That process, she concluded, put publishers, not servers, in the drivers’ seat:

[W]hen defendants caused the embedded Tweets to appear on their websites, their actions violated plaintiff’s exclusive display right; 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.

She also argued that Perfect 10 (which concerned Google’s image search) could be distinguished because in that case the “user made an active choice to click on an image before it was displayed.” But that was not a detail that the Ninth Circuit relied on in reaching its decision. The Ninth Circuit’s rule—which looks at who actually stores and serves the images for display—is far more sensible.

If this ruling is appealed (there would likely need to be further proceedings in the district court first), the Second Circuit will be asked to consider whether to follow Perfect 10 or Judge Forrest’s new rule. We hope that today’s ruling does not stand. If it did, it would threaten the ubiquitous practice of in-line linking that benefits millions of Internet users every day.