Retcon /ˈretkän/: Abbreviation of “retroactive continuity”; “Revise (an aspect of a fictional work) retrospectively, typically by introducing a piece of new information that imposes a different interpretation on previously described events.”
-- Oxford Dictionaries
Apparently, George Lucas is not the only party in California who can edit his own work after release in order to change aspects he does not like. While perhaps not as culturally significant as changing Star Wars: A New Hope so that Han Solo does not shoot first, Judge Tallman, writing for the Ninth Circuit, has now retconned the Court’s holding in the important Lenz v. Universal Music Group case to muddle what might have become a key protection for copyright owners. While refusing to rehear the motion upon which the original order was issued, the Court’s changes likely sow confusion for those seeking to enforce their intellectual property, while providing little additional comfort for ISPs and users who face frivolous take-down notices under the Digital Millennium Copyright Act.
As we described in a prior alert, in Lenz, the Ninth Circuit ruled that when submitting a take-down notice under the Digital Millennium Copyright Act (“DMCA”) – which requires an owner to have a good faith belief that the purportedly infringing material “is not authorized by the copyright owner, its agent, or the law” – copyright owners must first consider whether the use might be a fair use. The Court held that failing to do so exposes the copyright owner to potential liability for damages under 17 U.S.C. § 512(f), because “the statute unambiguously contemplates fair use as a use authorized by the law.” The Court then stated that the law requires copyright owners only to form a subjective, not objective, good-faith belief of infringement, and that so long such a belief was formed, courts cannot second-guess that conclusion.
The Court’s clarification that copyright owners bear this affirmative obligation of considering fair use was quite significant, because as we noted previously, the copyright take-down regime has become a massive undertaking by content owners, web hosts and users, and small shifts in burden can have large effects. The natural follow-up question that arose was: What does this affirmative obligation to consider fair use entail?
In the original decision by the Ninth Circuit, noting that it was “mindful of the pressing crush of voluminous infringing content that copyright holders face in a digital age,” the Court took up this follow—on question, at least in part, suggesting that the fair use analysis prior to a DMCA take-down notice “need not be searching or intensive” (emphasis added) and did not require an “investigation of the allegedly infringing content.” Indeed, the Court went even further and mused (in dicta) that computer algorithms might be employed to make a “first cut” fair use decision, supplemented by human review of “the minimal remaining content a computer program does not cull.”
Petitions for Rehearing
Both parties petitioned the full Court for rehearing. While Universal limited its argument to a jurisdictional (standing) issue, Lenz argued that the Court’s ruling allowed a content owner to effectively censor speech – especially speech by ordinary individuals – by permitting a subjective belief, even where unreasonably held, to act as a shield for frivolous take-down notices.
Google, Tumblr, Twitter and the owner of Wordpress submitted an amicus brief, arguing that unfounded or abusive take-down notices impose a significant cost on service providers as well as on free speech, and that determining liability “solely on the subjective knowledge of the notice sender … would lead to the illogical result that the more unreasonable a copyright holder is, the more legal leeway it has to send unfounded notices.” Two other entities, Public Knowledge and the Organization for Transformative Works – whose earlier brief had been cited by the Court for its reference to algorithms that can identify infringing content – also submitted an amicus brief, notably setting forth numerous ways in which take-down notices purportedly have been abused, including the use of “overbroad” matching algorithms.
Whatever it thought of the arguments of the parties and amici, the Ninth Circuit declined to rehear its decision affirming the district court’s denial of summary judgment, instead taking the opportunity to issue an amended decision. That amended decision deletes almost two full pages from the original, most importantly the paragraphs that set forth the type of fair use consideration that the Court originally indicated might suffice.
The decision also adds a footnote disagreeing with the dissenting judge, who believed that there was no triable issue of fact exist on fair use because Universal “did not specifically and expressly consider the fair-use elements.” The majority, however, held that the question of whether Universal’s fair use analysis was enough to form a subjective good faith belief that the video was infringed was a question for the jury.
Repercussions from the Amendment
What does all of this mean? As a threshold matter, it does not mean that in the Ninth Circuit a copyright owner now must undertake a searching and intensive fair use analysis before sending a demand. When a decision is amended, the prior decision may no longer be cited for the purposes of precedent, and thus the Ninth Circuit now simply has not defined the type of review that must be undertaken.
Given its insistence that the issue of a copyright’s owner’s subjective belief is for the finder of fact, the Court may simply believe that trial courts will be better situated to determine whether the fair use analysis was sufficient, in the context of particular facts. Of course, this alone likely tips more cases towards resolution by a jury, rather than on summary judgment. However, given that the Court also stated that merely forming a subjective belief is enough – regardless of whether it is proven correct – one imagines that so long as a content owner can show some proof of a decision-making process that includes fair use consideration, some cases will still be subject to summary judgment on this issue.
The amended decision does highlight that copyright owners cannot assume that a cursory fair use analysis will suffice. By not defining how careful the review must be, the court has left open the door for plaintiffs to argue that even though a fair use analysis occurred, it was insufficient.
To the extent that frivolous or abusive take-down notices constitute a significant problem for ISPs and their customers, as argued by the service provider amici, it is possible that the Ninth Circuit’s amended decision might make it more difficult for unscrupulous or careless parties to claim they formed a good-faith belief as to fair use when they have not done so. However, because the Court did not go as far as those parties wanted and left intact the subjective standard, and has not defined the level of review required by copyright owners to form that subjective opinion, it is not clear that the decision provides that much additional comfort for ISPs or users either.
On the other hand, perhaps the most troubling aspect of the revised decision, for copyright owners, is the removal of the Court’s language acknowledging the “pressing crush of voluminous infringing content that copyright holders face in a digital age,” and discussing the use of computers “for processing a plethora of content.” Perhaps the Court felt that requiring an owner to form a “subjective” opinion, while at the same time tacitly blessing computer algorithms to do the bulk of the decision-making, was confusing. Regardless, by deleting the language entirely it has muddled the issue even more. Will review using algorithms be adequate to form a subjective belief that there is no fair use, or are humans always required? The Ninth Circuit has now all but ensured that this issue will have to be litigated.
Our prior advisory suggested that copyright owners should have written procedures for identifying what content should be subject a takedown notice, and that those guidelines should expressly require a consideration of fair use. We also recommended that employees responsible for determining what material is infringing should receive information and training about fair use, and document their fair use determinations before issuing a takedown notice. This advice has not changed, although the latter steps have become more urgent in light of the Court’s revisions.
To the extent copyright owners employ automated screening algorithms to identify potentially infringing content in the wake of the Lenzamended decision, obviously it would be best if human reviewers double-check for fair use before sending take-down notices, which may be possible for owners that do not deal with a large volume of take-downs. However, where this is not practicable given a large volume of content, content owners must be aware that the risk of an unfavorable court decision on an improper take-down notice increases as the algorithms move beyond merely finding exact duplicates and are used to identify content that matches because of titles, key words or the like, and should employ second-level human review as appropriate.
Finally, as copyright owners begin to employ more advanced artificial intelligence tools to perform initial reviews, it will be interesting to see whether courts can ever become comfortable with treating highly-effective computer tools as analogous to providing the necessary “subjective” belief on the part of owners.
This seems unlikely in the near future. But if our prediction proves wrong, we can always retcon this advisory.