Posted by: Julia Ketchum
YouTube has become a huge internet success. Today, YouTube has over a billion users and everyday people across the globe watch hundreds of millions of hours of YouTube content. However, when YouTube was first created, it was unclear it would last. In 2007, Viacom sued YouTube since so much of its content was infringing copyrighted materials and Viacom argued YouTube did not do enough to monitor its own site. In response, unsure if the Digital Millennium Copyright Act (DMCA) would protect YouTube from liability, YouTube began working on Content ID in 2007. Content ID allowed YouTube to begin noticing copyrighted material on YouTube prior to getting notification from the copyright holder. It also could show to the court that YouTube was playing a more active role in monitoring its own site. Viacom Intern. v. YouTube ended with YouTube being protected by the DMCA Act, protecting YouTube from liability so long as YouTube followed the process outlined in 17 U.S. Code § 512 of the DMCA. Despite YouTube being protected with DMCA, YouTube still implements its Content ID today. In this post, we will examine how copyright holders typically use bots to generate their copyright claims under 17 U.S.C. § 512 and the shortfalls with that. Then we will move to an analysis of YouTube’s Content ID and how it faces the same shortfalls the copyright holder bots face. Then we will conclude how both systems have failed to use a Fair Use analysis and solutions for the future.
What is 17 U.S. Code § 512 of the DMCA?
This statute gives an online service provider a safe harbor, i.e. an immunity from liability in litigation, if the OSP (Online Service Provider) follows this statute appropriately. The provisions od 17 U.S.C. § 512(c) state an OSP will be immune from monetary relief for copyright infringement if it does not have actual knowledge of the copyright infringement, is not aware of facts in which the copyright infringement is apparent, or upon obtaining knowledge that infringement is going on, it removes the offending content from its site. The OSP also must have no monetary benefit from the infringing content. The OSP must also follow the steps after it receives a copyright notification from a copyright holder. The provisions of 17 U.S.C. § 512(c)(3) also lay out the elements of the copyright notification. The important factors for our purposes are that 1) the copyright holder must bring forth the claim to the OSP, alleging one of their exclusive copyright rights is being infringed and 2) the copyright holder must have a good faith belief that the alleged material is not authorized by the copyright holder or the law. Because this copyright notice is providing the OSP with knowledge that copyright infringement is allegedly occurring, under 17 U.S.C. § 512(c), the OSP needs to remove the offending content from its site or else risk losing its safe harbor immunity. There is also counter notice process as outlined in 17 U.S.C. § 512(g) to dispute the removal of the video. It is worth noting that under 17 U.S.C. § 512(g)(3), the person disputing the take down of her alleged infringing material must state her name, address, telephone number and consent to a federal jurisdiction court. This opens up the counter notice person for the high potential of being sued since any anonymity she had is lost in order to comply with the counter notice.
A look at the Copyright Holders rights and good faith under 17 U.S.C. 512(c)(3)
To file a copyright notification, one of the exclusive rights of a copyright holder must be violated. The exclusive rights of a copyright holder are set out in 17 U.S. Code § 106. This statute sets out six exclusive rights with one notable statement. The statute reads these rights are “subject to sections § 107 to § 122.” And 17 U.S.C. § 107 is the Limitations on exclusive rights: Fair Use. According to 17 U.S.C. § 106 and § 107, every copyright exclusive right is subject to a fair use exception. Also, in Lenz v. Universal Music Corporation, the court held under the good faith under 17 U.S.C. § 512(c)(3), a fair use analysis should be completed as part of the take down notice process.
How Do Copyright Holders Generate Notices and Is Fair Use Being Considered?
In most cases, copyrighter holders use bots to search for their copyrighted content and the bots are not able to account for fair use. Because the internet is such a vast landscape and YouTube alone has millions of videos, more added each day, a copyright holder is unlikely to have humans scouting for copyright infringement. Bots are programmed to search high and low for the copyrighted content without much ability to tell apart infringing material from the original material. In 2015, a bot hired by a third party filed a take-down notice to take down the copyright holder’s own site, the OSP complied as per 17 U.S.C. § 512(c) with the copyright holder having to file a counter notice. A FOX Broadcasting bot once flagged a video made well before the copyrighted material aired, which the copyrighted material actually stole and used for its show. Then there are the stories of bots flagging items that a human would instantly see as fair use. A Universal Studios bot (hired by a third party) asked Google to flag the ImdB (Internet Movie Database) page for Fast and the Furious 7. The Internet Movie Database would be protected under fair use since fair use protects “for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright.” The Internet Movie Database would likely be considered criticism, comment, and/or news reporting since it collects information on movies in a searchable database, providing reviews for the movie as well as links to articles about the movie. A copyright bot for a local news station filed a notice with YouTube (and YouTube complied) to take down a live broadcast from a NASA rover. Clearly, NASA (National Aeronautics and Space Administration) videos would be considered news reporting and for teaching or research purposes. Copyright bots continue to fail to consider fair use.
What is YouTube Content ID
Since YouTube’s lawsuit against Viacom, YouTube has wanted to take a more proactive response in detecting copyright infringement. Even after YouTube prevailed and was told it was protected from liability under DMCA so long as it followed 17 U.S. Code § 512, YouTube has continued its Content ID program. To be clear, the DMCA does not require YouTube (or any OSP) to proactively search for infringing content. The DMCA requires for OSPs to respond to copyright holder claims of infringement or respond if the OSP has actual knowledge of the infringement. However, more and more OSPs are following YouTube’s example and are moving beyond the scope of the DMCA to find infringing material.
YouTube Content ID is fully explained in this video made by YouTube on YouTube’s site. Basically, copyright holders give YouTube a copy of their copyrighted material to put in a data base YouTube maintains as ‘reference files.’ YouTube states in the video “Every time you upload a video, we quickly compare it to the reference files…looking for a match.” It looks for matches by comparing audio, video, partial matches, and can even can make a match if the quality of the video is very different from the quality of the reference file. After a match is identified, YouTube does whatever the copyright holder wants them to do. Copyright holders at some point tell YouTube which option they want: block the video, do nothing to the video, or start making money off the video by diverting ad revenue. Just like the DMCA, there is an appeals process, which requires the user to fill out forms, again, potentially losing anonymity.
Is Content ID Able to Account for Fair Use
Many of the same problems copyright holder bots have been repeated in Content ID. Content ID is unable to see fair use exceptions. Since Content ID is comparing matches, it seems like it was never programmed to account for Fair Use. There are various examples of Content ID flagging matches and taking down videos that would clearly be fair use. In 2012, Content ID flagged and took down First Lady Michelle Obama’s speech from YouTube. YouTube live streamed the speech as it was given, but somehow the video was flagged by Content ID once the stream ended. Following the copyright holder’s request to block the video, the video was blocked. Content ID has also flagged a YouTuber for singing Christmas Carols that belong in the public domain. A popular YouTuber was singing Silent Night and Content ID matched his song with various recording companies who had famous people sing the same song at some point. Despite the obvious knowledge that a song over 150 years old is not protected by copyright, Content ID followed the recording companies’ suggestion to block it. After an appeals process, the user got to have his video put up again. Another popular YouTube personality, Angry Joe, who has over two million subscribers, has encountered issues with Content ID. Angry Joe does reviews of video games and would likely be covered under criticism or news reporting of fair use. However, in his video from 2013, he rants about sixty-two of his video reviews were suddenly flagged overnight as infringing. He got flagged for an interview he did with a video game company since the video game company apparently flagged it under a reference file, despite the obvious fact that Angry Joe was the reporter in the interview and should have some right to his own interview. He got flagged for music in the background of a clip he showed during a criticism of a game. It is worth noting that Angry Joe does not show much of the video games he reviews. And most would agree to review content, you can show the thing you are reviewing. Three years later, Angry Joe is still having the same problem with Content ID. In 2016, he released a video criticizing YouTube Heroes. After criticizing YouTube, Content ID matched his video and tried to block it, unleashing another Angry Joe rant. It is clear Content ID is failing to account for fair use as well.
Content ID Does More Than Take Down: It Diverts Money
If anything, Content ID is worse than the copyright bots since Content ID allows copyright holders to earn money off allegedly infringing material with no fair use examination. As stated above, after Content ID makes a match, Content ID follows what the copyright holder wants to do and one of the options is to make money off the video. After a match is identified, the copyright holder can choose to divert ad revenue from the alleged infringing video to themselves. One example is the Wired story of Masae Anela. Masae Anela is a ‘let’s player.’ A let’s play when someone streams an entire video game, adding their own commentary. There is another post on Cyberbeartracks available here about the legal status of let’s plays. So far, there has not been a definitive ruling on the legal status of let’s plays. However, as the Cyberbear Tracks post argues, the new commentary could support a transformative purpose that would push the Let’s Play into Fair Use ground. However, in the Masae Anela case, the Content ID system merely matches the audio/video against the reference file and flags her video. Nintendo, the copyright holder in this case, determines it wants a cut of her ad revenue. YouTube Content Providers like her make money off ad revenue; it is how they support themselves to be a content provider full-time. And somehow, the fair use test is ignored and the burden is put on Masae Anela to bring it up in her defense. Guilty until proven innocent. However, like most people when facing a notice, she is unsure what to do. In the end, she accepts Nintendo taking a cut of her money and continues on. The introduction of money into the problem presents more problems. If Masae had appealed Nintendo’s decision, she would have opened herself up for a copyright infringement suit. If she wins the appeal, her ad revenue from the complaint to the appeal is lost forever. Money diversion happens immediately after the copyright holder requests it, regardless of the strength of the defense. And with this system in place, if you are a copyright holder, why not go for the money every time? There does not seem to be any punishment for copyright holders making frivolous claims against YouTube Content Providers.
In conclusion, copyright bots and YouTube Content ID fail to follow the law. Every exclusive copyright is subject to fair use and yet bots are unable to identify fair use. This problem has been consistent and will remain consistent until either bots are programmed better or humans take an active role in looking at content prior to the take-down notice being issued.