Wednesday, 23 November 2011 15:22
Ever since YouTube first introduced its automated “Content ID” copyright filtering system, the potential has existed for tremendous abuse. Over the last couple years, evidence has been mounting that YouTube’s Content ID system is in fact being systematically used to falsely claim and monetize videos the claimant has no right to, and to block videos with no recourse for the user, even if they are likely fair use.
All of this is done with YouTube’s direct knowledge and consent, having designed a system that is wide open for abuse and lacking in any effective and transparent dispute or appeals process. This page attempts to document this problem in hopes that Google/YouTube will eventually be pressured into changing its policies to rectify this situation.
A. Introduction: Three Fundamental Flaws with the Content ID System
In order to understand the problem of copyright fraud enabled by YouTube’s Content ID system, it is first necessary to understand the three fundamental flaws in its design which enable this abuse.
1. The Content ID program apparently requires no proof of copyright ownership
YouTube’s Content ID system works by allowing copyright owners to upload digital copies of video or audio works, which YouTube’s servers use to create a digital “fingerprint,” against which all other videos on the site are scanned. If even a portion of another video matches the sample in either its visual or audio content, the video is flagged as containing that copyrighted content. From there, the copyright claimant can choose to either block, allow, or “monetize” matched videos. Monetization is done by allowing YouTube to run ads next to the a video, from which the copyright claimant receives a cut of the ad revenue.
While it seems obvious that a system which allows alleged copyright owners to upload any audio/visual work and claim copyright ownership over that work should at minimum require that person to provide some documentation or proof that they own the copyright to each work they claim, anecdotal evidence suggests that no such proof is in fact required. The sheer number of Content ID claims involving content which the claimant could not possibly have any copyright interest in, indicates that either no proof of copyright ownership is required at all, or at least that YouTube does very little to verify this and copyright ownership is easily faked.
The impact of this flaw is that anyone who manages to gain admission to the Content ID program can upload any content they want into the system, which it then flags as belonging to them. They can then block or monetize any video they want, regardless of whether they really own any copyright interest in it. As a result, there is nothing to prevent any Content ID partner from uploading a copy of the latest popular viral video and claiming it as there own, allowing them to hijack the ad revenue from that video, which can be substantial.
2. Content ID identifications are notoriously inaccurate
While YouTube claims the Content ID system results in very few false positives, experience suggests these matches are highly inaccurate and incapable of considering the context of the material in question. Many works are outright misidentified. In other cases, the specific work is correctly identified, but matched incorrectly. For example, if a Content ID partner makes video game reviews which include cutscenes from a popular videogame, Content ID might attribute all other videos using cutscenes from that game to the other reviewer. Likewise if a song by one artist uses royalty free music loops from something like GarageBand or Final Cut, and another song by a different artist uses those same loops, Content ID may identify the second song as matching the first–even though the only elements those songs have in common are in the public domain.
3. The Content ID dispute process is ineffective and gives copyright claimants the ability to unilaterally “confirm” their claim with no further recourse for the uploader
The Content ID system includes a supposed “dispute” process, wherein a user who believes his video has been incorrectly flagged by Content ID can use a simple webform to dispute the Content ID match for one of three reasons (1) misidentification, (2) license to use the material in question, and (3) fair use. YouTube describes this dispute process as being a sort of front-end buffer to the notice and counter-notice process established by the Digital Millennium Copyright Act (17 USC § 512). According to YouTube, once the user files a dispute, the video is automatically restored. If the copyright claimant wishes to have the video blocked again, they must file a DMCA takedown notice. If the user still disagrees, they can then file a DMCA counter-notice and get the video restored. At that point, if the copyright claimant still objects to the video, they must file a lawsuit seeking an injunction.
If this was the way the dispute process actually worked, the impact of the first two problems would be minimal, as false identifications could be easily corrected by filling out a simple online form. As it is however, this is not the way the dispute process works in practice. In reality, after the original uploader files a dipsute, YouTube allows the Content ID claimant to simply “confirm” their claim to the video, allowing them to either permanently block or monetize the video with no further recourse for the uploader. The user is then met with a message saying “All content owners have reviewed your video and confirmed their claims to some or all of its content.” After that, there is nothing the uploader can do to fight the copyright claim on their video. Further Content ID disputes are not allowed, and neither can they file a DMCA counter-notice because no DMCA takedown notice has been filed. This conundrum is illustrated by the following diagram:
The upshot of this is that Content ID allows claimants to unilaterally reject any dispute filed against their own copyright claim, essentially making them the judge of their own case. Many, if not most, entities that use the Content ID system make it their policy to blanketly confirm their claims against all disputes that are filed, and don’t even bother to evaluate whether individual disputes have merit or not. As a result, the Content ID dispute process as it exists now is simply a joke, and provides no effective means to appeal false copyright claims against a user’s videos. This leaves the system wide-open for abuse, with little to no accountability for those who seek to use the Content ID system for nefarious purposes. Read more…