Copyright Controversy over Bird Songs Highlights Flaws of Automated Content ID Systems
Why was a YouTube video containing no music flagged as containing a musical composition and hit with a copyright infringement claim? This question arises after a British vegan and YouTube user, “eeplox,” uploaded a video about foraging and making a wild salad from mustard leaves and dandelions. The video contained no music, but the recording’s background noise featured bird sounds. (No, not “Mr. Tambourine Man,” but actual birds chirping.) YouTube sent the user a notice that it had identified the video as containing a copyrighted musical composition belonging to Rumblefish. It appears that the copyright claims process had two points of failure in this case.
First, YouTube uses a fully automated system to identify musical content in a given video and associate it with registered users in commercial partnerships, like Rumblefish. If this content ID system classifies a video as making improper use of copyrighted content, YouTube provides notice to the uploading user by sending a message informing them that the video contains copyrighted music belonging to another entity. The user then has the opportunity to file a dispute on the infringement claim, as eeplox did in this case. Once a dispute is filed, notice is sent to the entity that YouTube believes owns the copyrighted content, here giving Rumblefish an opportunity to review the claim.
However, a second failure occurred when a representative from Rumblefish reinstated the claim, asserting ownership of a nonexistent musical composition on the video. YouTube planted advertisements on the video in lieu of taking it offline, and Rumblefish seemingly profited from it for a time.
Fortunately for eeplox, his story gained significant web attention, and the claim against his video was eventually removed, but only after the Rumblefish CEO, Paul Anthony, got involved. Anthony first took to Reddit to explain what happened and deflect his company’s responsibility for the series of errors. His formal statement on the subject seemed to downplay the mistake, casting this as a one-off incident, and he placed most of the blame on YouTube’s Content ID technology, but pledged to improve his company’s dispute review process. Nevertheless, this offers little consolation for users that fall victim to similarly bogus infringement claims over recorded noises from motorcycle engines or remote-controlled plane propellers.
While incorrect accusations of copyright infringement seem like harmless mistakes or mere inconveniences for YouTube’s users, it can have significant consequences. If the content ID system is triggered, advertisements on a user’s entire YouTube channel are disabled, so affected users cannot make money from their uploaded content until the dispute on the infringement claim is answered. Thus, errors like this must be avoided or, at the very least, be made easier to remedy. Currently, YouTube’s self-defensive copyright policy places the onus on users to dispute mistaken infringement claims. In cases where companies mistakenly reinstate infringement claims, users are seemingly left entirely at their mercy to resolve the issue. Certainly, the use of audio scanning technology to check for copyright infringement is reasonable and appropriate. But because the current content ID system is programmed so sensitively and characterized by such a high level of automation, there must be a human check to ensure that average users aren’t harmed by YouTube’s efforts to ensure compliance with the DMCA. The status quo allows companies to recklessly assert rights over videos they don’t own.