A three-judge panel of the Ninth Circuit handed down its latest decision on the scope of the optional safe harbor for web hosting services under Section 512(c) of the Digital Millennium Copyright Act. The April 7 decision in the case, Mavrix Photographs, LLC v. LiveJournal, Inc., limits the availability of safe harbor protection for social media platforms and other sites that use moderators to review user-submitted posts in copyright cases, even where the website has processes in place for expeditiously removing materials identified in DMCA takedown notices. And it will likely have broad implications on what screening procedures, if any, a service provider adopts for user-generated posts and material.
The case concerns 20 photographs posted to a popular community for celebrity news called “Oh No They Didn’t!”, hosted by the social media platform LiveJournal. ONTD users submit proposed articles, and moderators decide whether to post them publicly. The moderators themselves are typically unpaid volunteers. LiveJournal also has notice and takedown procedures that copyright holders may use to notify LiveJournal of alleged infringements on ONTD or other communities. Rather than use these procedures to have the photos removed, however, Mavrix—a celebrity photography agency that owns the copyrights for several photographs of Beyoncé Knowles—filed suit against LiveJournal.
The district court ruled in favor of LiveJournal. On summary judgment, it held that the Section 512(c) safe harbor shielded LiveJournal from liability because, although moderators screened all ONTD posts, the posts were made at the direction of the user. Moreover, with respect to discovery, the lower court denied Mavrix’s motion to compel disclosure of the identity of ONTD moderators, based on the moderators’ First Amendment right to anonymous internet speech.
On appeal, the Ninth Circuit reversed the summary judgment holding, vacated the discovery order, and remanded for further proceedings.
To be eligible for the Section 512(c) safe harbor for hosting platforms, a service provider must, among other things, show that the materials at issue are stored “at the direction of a user.” The Ninth Circuit held that this threshold inquiry “turns on the role of the moderators in screening and posting users’ submissions and whether their acts may be attributed to LiveJournal,” the service provider. The Mavrix Court also found that the district court erred by not applying common law principles of agency to this determination, and it adopted the factor-based analysis appearing in the 1989 Supreme Court decision of Community for Creative Non-Violence v. Reid.
In Reid, the Supreme Court applied agency principles to determine whether material constitutes a work made for hire under the Copyright Act. The Reid framework is intensively factual and requires consideration of several factors, including but not limited to the skill required; the location of the work; the extent to which the individual has discretion over when and how long to work; and the method of payment, if any.
Ultimately, the Mavrix Court held that the record raised genuine issues of material fact regarding the level of control LiveJournal exercised over ONTD moderators, and it remanded for further fact-finding on this point. In particular, the Ninth Circuit marked a distinction between activities directed toward “enhancing the accessibility of the posts,” i.e. screening for harassment or pornography, and more extensive review: “The fact finder should determine whether LiveJournal’s manual, substantive review process went beyond the automatic processes we have approved as accessibility-enhancing activities such that the posts were still at the direction of the user.”
Last April, in BWP Media USA Inc. v. Clarity Digital Group., LLC, the Tenth Circuit also considered the scope of the Section 512(c) safe harbor with respect to materials stored “at the direction of a user.” There, the BWP Court tackled the question of whether contractors can qualify as “users” for purposes of Section 512(c). The Tenth Circuit held that they could, ruling that a hosting service may qualify for safe harbor under the DMCA for the conduct of paid independent contractors. The BWP Court suggested that safe harbor protection may extend even to employee conduct, under a plain meaning reading of the statute: “simply because someone is an employee does not automatically disqualify him as a ‘user’ under § 512.”
With Mavrix, the Ninth Circuit raises a potential conflict with the Tenth Circuit’s BWP decision. Indeed, in a footnote addressing the Tenth Circuit’s decision, the Mavrix Court appears to announce a circuit split: “To the extent that BWP’s holding contradicts our case law that common law principles of agency apply to the DMCA such that a service provider is liable for the acts of its agents, including its employees, we reject it.” While it is still too early to tell, given the important role the DMCA plays in the operations of national internet service providers, Mavrix may soon become the first case before the Supreme Court to squarely address the DMCA’s safe harbor provisions.
Beyond vacating and remanding the district court’s decision on the threshold question of agency, the Ninth Circuit offered guidance on additional conditions for Section 512(c) safe harbor that it determined “may be contested on remand.” Once “an internet service provider shows that the infringing material was posted ‘at the direction of the user,’ it must then also show that (1) it lacked actual or red flag knowledge of the infringing material; and (2) it did not financially benefit from infringements that it had the right and ability to control.” The failure of a service provider to establish either of these conditions defeats its eligibility for the safe harbor. On both points, the Mavrix Court narrows the availability of the safe harbor, at least with respect to service providers that use moderators, and it also suggests that service providers may have to incur significant additional discovery costs to litigate the safe harbor issue.
With respect to actual or red flag knowledge, the Ninth Circuit acknowledged that Mavrix, in failing to provide notice of alleged infringements to LiveJournal before suit, had foregone its strongest evidence of knowledge. But the Court then went on to offer Mavrix a road map of the types of activities that might defeat LiveJournal’s assertion that it lacks knowledge of infringement. For instance, the Court proposed that Mavrix should be able to compel disclosure of the identities of individual moderators, in order to depose them about their subjective knowledge of infringement. Setting aside the First Amendment’s protection of a moderator’s right to anonymous internet speech, this inquiry could substantially increase the cost of discovery for service providers seeking safe harbor protection. Similarly, the Ninth Circuit suggested that the presence of a watermark on the photograph may constitute knowledge sufficient to defeat the safe harbor: “To determine whether LiveJournal had red flag knowledge, the fact finder should assess if it would be objectively obvious to a reasonable person that material bearing a generic watermark or a watermark referring to a service provider’s website was infringing.”
The upshot is that a court applying Mavrix may well attribute red flag knowledge to a hosting service that uses moderators based merely on the existence of images containing watermarks on its site. That leaves a service provider seeking Section 512(c) safe harbor protection with two distasteful options: either the site does not use moderators at all, or it affirmatively screens posts for watermarks or other indicators of potential infringement. The latter course is not only unduly burdensome, it directly contradicts Section 512(m)’s promise that a service provider need not affirmatively monitor its service for allegedly infringing activity to be eligible for safe harbor under the DMCA. And the former means sacrificing efforts to quality control or curate the materials on a platform.
Meanwhile, the Mavrix Court’s discussion of what evidence is necessary for LiveJournal to show that it did not financially benefit from infringements that it had the right and ability to control further calls into question the availability of the safe harbor to sites with moderators. Here, the Ninth Circuit agreed with the district court’s conclusion that the right and ability to control requires “something more” than the mere ability to remove or block access to allegedly infringing materials. But it also rejected the notion that LiveJournal’s screening practices, i.e. its “extensive review process, infringement list, and blocker tool,” could not satisfy this showing, and it remanded that issue to the fact finder.
The Court also distinguished the DMCA’s direct financial benefit inquiry, § 512(c)(1)(B), from both the threshold determination under § 512(c)(1) and the knowledge showing under § 512(c)(1)(A). For the financial benefit inquiry, the Ninth Circuit noted, the “fact finder should consider the service provider’s general practices, not its conduct with respect to the specific infringements.” And it advised that a jury might reasonably conclude on remand that the mere presence of allegedly infringing material on an ad-supported website—with no causal link to the specific material a copyright plaintiff claims to be infringing—might suffice to establish a financial benefit and defeat safe harbor.
By focusing on a service provider’s “general practices,” the DMCA’s direct financial benefit analysis under Mavrix is more onerous than anything required under the substantive standards for secondary copyright liability. Indeed, just earlier this year, the Ninth Circuit in Perfect 10 v. Giganews reaffirmed its view that a copyright plaintiff “must demonstrate a causal link between the infringing activities and a financial benefit” to a defendant service provider to establish a claim of vicarious liability for copyright infringement. Thus, the Mavrix decision suggests that making out a DMCA defense requires evidence that is qualitatively different from that which a copyright defendant would rely upon to defeat a substantive claim of infringement: an infringement claim requires a copyright plaintiff to make affirmative showings tied to specific asserted works, while the DMCA safe harbor considers evidence that may bear no relationship to any works that a plaintiff asserts.
Even where a website, like LiveJournal here, has procedures for expeditiously removing allegedly infringing user-generated materials identified on takedown notices, Mavrix teaches that a service provider may still lose safe harbor protection based on its efforts to curate materials available on its platform. This ruling incentivizes platforms seeking to avail themselves of the safe harbor to exercise less oversight of materials that users submit for posting. That is particularly true in light of the impracticability of having a platform, which lacks both information about the ownership of posted materials and competence to make judgments about their legal status, to screen materials based on infringement.
Indeed, the Mavrix ruling demonstrates that the Section 512(c) safe harbor may not be that safe after all. In many instances, a service provider in the context of a litigation may be better off not invoking it. That way the burden remains where it should: on the plaintiff to establish the elements for its affirmative claims based on the substantive standards for infringement, rather than on the defendant to prove entitlement to the safe harbor.