With over one billion websites on the Internet, and 211 million items of online content created every minute, it should come as no surprise that content curation is one of the hottest trends in the Internet industry. We are overwhelmed with online content, and we increasingly rely on others to separate the good from the bad so that we can make more efficient use of our time spent surfing the web.

Consistent with this trend, many websites that host user-generated content are now focused on filtering out content that is awful, duplicative, off-topic, or otherwise of little interest to site visitors. And these sites often find that humans—typically passionate volunteers from the sites’ user communities—are better than algorithms at sorting the wheat from the chaff.

Of course, any website that deals with user-generated content needs to consider potential copyright liability arising from such content. We’ve discussed in past Socially Aware blog posts the critical importance of Section 512(c) of the Digital Millennium Copyright Act (the DMCA) to the success of YouTube, Facebook and other online platforms that host user-generated content. By providing online service providers with immunity from monetary damages in connection with the hosting of content at the direction of users, Section 512(c) has fueled the growth of the U.S. Internet industry.

Nevertheless, questions have persisted as to whether a website operator’s use of employees or volunteers to review and filter user-generated content could deprive the operator of protection under the Section 512(c) safe harbor. A new Ninth Circuit decision, Mavrix Photographs, LLC v. LiveJournal, Inc., tackles this subject head on, concluding—to the alarm of many online service providers—that a company’s use of volunteer moderators to curate content hosted on its website resulted in triable issues of fact as to whether the company qualifies for DMCA safe harbor protection to the extent that such content infringes third-party copyrights.

LiveJournal and the Oh No They Didn’t! Community Journal

The defendant in the case, LiveJournal, operates a social media platform on which a user can, among other things, create a “community journal” focused on a particular topic or theme. As the name suggests, a community journal allows multiple users to submit content to the journal. The creator or administrator of a community journal decides whether to enable moderation, to have no moderation, or to employ a hybrid moderation model (e.g., whitelisting certain domains while requiring moderation for others).

The dispute involved a moderated community journal focused on celebrity gossip and pop culture, called “Oh No They Didn’t!” (ONTD). Initially run entirely by volunteers, ONTD created rules to govern user submissions and posted comments; these rules included a requirement that users incorporate any referenced articles and photos in their submissions (rather than linking to such items), and a prohibition on the submission of old gossip and outdated photos. ONTD also adopted a “blacklist” of third-party sources from which users were prohibited from copying materials for inclusion in their submissions.

By 2010, ONTD was generating 52 million page views a month, making it the most popular journal on the LiveJournal platform. To maintain this success, ONTD relied on volunteer moderators (who review user-submitted items to ensure compliance with ONTD’s rules) and maintainers (who review and delete posts, and have the authority to remove users and even moderators from the ONTD community).

That same year, for the first time, LiveJournal became directly involved in the operation of ONTD. In particular, it hired one of ONTD’s volunteer moderators—Brendan Delzer—to serve as the journal’s full-time “primary leader.” In this new role, Delzer instructed ONTD moderators regarding their own review and selection of content, and monitored and evaluated moderators, removing underperformers; Delzer also continued to review and select user-submitted content for posting to ONTD. By making Delzer an employee, LiveJournal sought to “take over” ONTD, in order to further grow the journal and run ads against the journal’s content.

The Parties’ Dispute

The plaintiff, Mavrix, is a company that specializes in candid photographs of celebrities; it generates revenue by selling its photos to magazines. Between 2010 and 2014, 20 celebrity photographs owned by the plaintiff were allegedly posted to ONTD in seven separate posts, all without Mavrix’s authorization. Some of these photos incorporated either a generic watermark or a Mavrix-specific watermark.

In 2014, in lieu of sending a DMCA takedown notice, Mavrix sued LiveJournal for copyright infringement in connection with the 20 photographs. Upon learning of the suit, LiveJournal removed the photographs from its site and terminated the LiveJournal account holders who had posted the photographs. LiveJournal ultimately moved for summary judgment on the basis that it was shielded from copyright liability under the Section 512(c) safe harbor.

Mavrix opposed LiveJournal’s motion and filed its own motion asserting that LiveJournal was disqualified from the Section 512(c) safe harbor because, as a result of the activities of ONTD’s volunteer moderators acting as LiveJournal’s agents:

  • The photographs had not been hosted on the LiveJournal site “at the direction of a user,” as required under Section 512(c);
  • LiveJournal had actual or at least red flag knowledge of the photographs through the ONTD moderators; and
  • LiveJournal received a financial benefit directly attributable to the allegedly infringing activity, where it had the right to and ability to control such activity.

The district court granted LiveJournal’s summary judgment motion, finding that ONTD moderators constituted independent third-party users, not agents of LiveJournal. Mavrix appealed, and the Ninth Circuit reversed.

Storage at Whose Direction?

As noted above, the Section 512(c) safe harbor is available only where content is stored by an online service provider “at the direction of the user.” In reaching its decision, the Ninth Circuit focused on whether LiveJournal’s use of moderators to pre-screen user-submitted content meant that the Mavrix photos were posted at LiveJournal’s direction, rather than at the direction of the users who submitted such photos to ONTD.

After reviewing the sparse case law on the subject, the Ninth Circuit held that the common law of agency applies in determining whether the acts of the ONTD moderators are attributable to LiveJournal, and that genuine issues of material fact existed as to whether the moderators are LiveJournal’s agents.

In reaching this conclusion, the Ninth Circuit examined the common law agency concepts of actual and apparent authority and the level of control of a principal over the agent, and found:

  • Regarding actual authority (which arises where the principal assents to the agent taking action on the principal’s behalf), Mavrix had presented evidence that LiveJournal “gave its moderators explicit and varying levels of authority to screen posts” and provided “express directions about their screening functions.” Unlike other online platforms where users may independently post content, “LiveJournal relies on moderators as an integral part of its screening and posting business model.”
  • Regarding apparent authority (which arises when a third party reasonably believes that a person has authority to act with legal consequences on behalf of another), Mavrix had presented evidence that ONTD users may have reasonably believed that the journal’s moderators had authority to act for LiveJournal.
  • Regarding the level of control, Mavrix presented evidence showing that “LiveJournal maintains significant control over ONTD and its moderators”; that Delzer—a LiveJournal employee—had engaged in “substantive supervision” of ONTD moderators, and had exercised the power to select and remove moderators; and that LiveJournal had “ratified” the ONTD-developed rules that moderators relied on in screening user submissions.

Based on its analysis of actual authority, apparent authority and control principles, the Ninth Circuit held, “From the evidence currently in the record, reasonable jurors could conclude that an agency relationship existed” between LiveJournal and ONTD’s volunteer moderators.

The Ninth Circuit next turned to a related issue: If the fact finder were to determine that the ONTD moderators are agents of LiveJournal, the fact finder would then need to determine whether Mavrix’s photos “were indeed posted at the direction of the users in light of the moderators’ role in screening and posting” the photos. The court noted that posts are at the direction of the user if the online service provider (1) “played no role in posting them on its site” or (2) “carried out activities that were ‘narrowly directed’ towards enhancing the accessibility of the posts.”

Regarding this second item, the Ninth Circuit observed that activities such as the automatic reformatting of posts or the manual screening of posts “for infringement or other harmful materials like pornography” can be “accessibility-enhancing,” and thus do not necessarily deprive an online service provider of protection under Section 512(c). But, if ONTD’s moderators are determined to be LiveJournal’s agents, the fact finder would then need to decide “whether the moderators’ acts were merely accessibility-enhancing activities or whether instead their extensive, manual, and substantive activities went beyond the automatic and limited manual activities we have approved as accessibility-enhancing.” In this regard, the Ninth Circuit noted that the ONTD moderators manually review the substance of user-submitted items to ensure that they contain new and exciting gossip and that only one-third of all submitted items are ultimately approved for posting.

Lack of Knowledge

Having provided guidance to the lower court on the threshold requirement of whether the Mavrix photos were stored at the direction of users, the Ninth Circuit turned to another Section 512(c) safe harbor requirement disputed by the parties: Whether LiveJournal lacked actual and red flag knowledge of such users’ infringing activities. If, on remand, the fact finder were to determine that the photos were stored by LiveJournal at the direction of users (rather than at LiveJournal’s own direction via its agents), LiveJournal would then need to establish that it had neither actual nor red flag knowledge that the photos were infringing.

The actual knowledge inquiry looks at whether an online service provider has subjective knowledge of the specific infringing activity at issue. The district court had held that actual knowledge can only be conferred through a DMCA-compliant takedown notice received by a service provider from the copyright owner; because Mavrix had never sent such a notice, LiveJournal necessarily lacked actual knowledge of its users’ infringing activities in connection with the Mavrix photos.

Sending shockwaves through the online service provider community, the Ninth Circuit rejected the district court’s approach, holding that a fact finder may determine that LiveJournal—either directly or through its agents, the volunteer moderators—in fact possessed subjective (and thus actual) knowledge of the infringing posts, even in the absence of a takedown notice from Mavrix.

Further, even if LiveJournal can establish on remand that it lacked actual knowledge, it must also show that it did not have red flag knowledge of the infringing posts. Red flag knowledge exists where an online service provider is aware of facts that would have made the specific infringement at issue objectively obvious to a reasonable person. Such infringement must be “immediately apparent to a non-expert.”

The Ninth Circuit noted that some of the posted Mavrix photos contained either a generic watermark or a Mavrix-specific watermark, and that, on remand, the fact finder will need to determine whether it would have been objectively obvious to a reasonable person that photos bearing such watermarks were infringing.

Lack of Financial Benefit and the Right and Ability to Control

The Ninth Circuit then turned to the final Section 512(c) safe harbor requirement at issue: Even if LiveJournal were to establish on remand that the Mavrix photos were stored at the direction of users, and even if it were able to establish that it lacked actual or red flag knowledge that the photos were infringing, it then must show that it did not financially benefit from infringing activities that it had the right and ability to control.

The Ninth Circuit stated that the right and ability to control component of this final requirement involves “something more” than the mere ability of the service provider to remove or block access to materials posted on its website. This “something more” exists where the service provider exerts “high levels of control” over the activities of users. The court observed that, on remand, the fact finder should take into account the following in assessing whether LiveJournal had a right and ability to control the infringing posts:

  • LiveJournal ratified ONTD rules that instruct users on the substance of their posts;
  • ONTD moderators screened the substantive content of user submissions;
  • Nearly two-thirds of posts submitted by ONTD users were rejected by ONTD moderators, including on substantive grounds;
  • ONTD prohibited users from submitting content taken from third-party content sources identified on a “blacklist” maintained by ONTD; and
  • LiveJournal implemented a mechanism to automatically block any submitted items originating from one of the sources on the ONTD blacklist.

The court concluded that “the fact finder must assess whether LiveJournal’s extensive review process, infringement list, and blocker tool constituted high levels of control to show ‘something more’.”

Finally, the Ninth Circuit noted that LiveJournal would need to establish that it did not derive a financial benefit from infringement that it had the right and ability to control. A safe harbor-disqualifying financial benefit “need not be substantial or a large portion of the service provider’s revenue,” and could be based on a broad availability of infringing materials attracting advertisers to the service provider’s site. On remand, the fact finder would need to consider the following:

  • LiveJournal’s receipt of revenue from advertising based on the number of ONTD page views; and
  • Evidence presented by Mavrix (but contested by LiveJournal) showing that approximately 84% of posts on ONTD contain infringing material

In order for LiveJournal to prevail on its safe harbor defense, the fact finder would need to determine whether LiveJournal met its burden of establishing that it did not financially benefit from infringement or that, if it did so benefit, it lacked the right and ability to control such infringement.

Concluding Thoughts

The Ninth Circuit’s Mavrix opinion may be the most significant DMCA safe harbor decision since the Second Circuit’s landmark Viacom Int’l, Inc. v. YouTube, Inc. ruling in 2012. On numerous fronts—including what constitutes “at the direction of a user,” what constitutes “actual knowledge,” what constitutes “red flag knowledge,” what constitutes a “right and ability to control” and what constitutes a “financial benefit”— the Mavrix decision appears to go against the tide of generally pro-service provider decisions on these issues over the past two decades.

Regardless of what the district court ultimately decides on remand, Mavrix is a wake-up call to website operators and other online service providers who rely on volunteers to screen, review or curate user-generated content. Such operators and service providers will want to carefully study the Ninth Circuit’s decision to determine whether they should take steps to modify their current practices regarding user-generated content. And, although Mavrix focuses on volunteers, not employees, companies should keep in mind that employees engaged in screening, reviewing or curating user-generated content are even more likely than volunteers to qualify as company agents, and thus may raise some of the same DMCA safe harbor concerns discussed in the Mavrix decision.