Do Copyright Holders Get a Free Pass to Identify Alleged Infringers?
by Paul Alan Levy
Suppose that, on June 4, 2022 the owner of a China-based Twitter account were to post this 2005 photo of the Gate of Heavenly Peace, with text calling for remembrance of the martyrs of Tiananmen Square.
Suppose further that, the following week, a previously unknown company based in China and claiming ownership of the copyright in the 2005 version of the iconic painting of Mao’s bust that sits over the main entrance to the Gate (a new version is hung annually), or falsely claiming copyright in the photo, were to issue a DMCA takedown to Twitter, followed by a subpoena demanding to know the identity of the account holder, claiming that the information is needed to enforce that copyright. Would Twitter be legally obligated to hand over the name of the account holder without delay, knowing that the likely result could be an all-expenses paid trip for the Twitter user to a re-education camp in Xinjiang?
That is what Big Copyright, represented by the Copyright Alliance and by a coalition of trade associations of professional photographers, argues in a pair of amicus briefs that they have filed in the controversy over Twitter postings by CallMeMoneyBags. The case presents the latest challenge in the Northern District of California to the well established rule, first established in 2001 in Dendrite International v. Doe, previously followed in that district as well as in federal and state courts across the country, that a would-be plaintiff claiming that its rights have been violated by anonymous online speech has to make a showing, supported by both legal argument and an evidentiary showing, that it has a tenable claim before it gets to identify the anonymous speaker and thus obtain the ability to serve the speaker with a summons and complaint and litigate its claim to a conclusion, and that assuming the plaintiff has made these showings, the court has to balance the right to enforce substantial claims through litigation against the prospective costs of breaching the right to speak anonymously.
The Facts of the Case
The case arises in somewhat peculiar circumstances. Over a period of a few days in October 2020, a Twitter user using the handle “CallMeMoneyBags,” who specialized in tweets about private equity figures, posted a series of photographs of nubile women to which he appended texts suggesting that Brian Sheth, a private equity billionaire, was now investing his wealth in such women (the tweets and photos appear at pages 3 to 5 here). A mysterious entity named Bayside Advisory, LLC, which appears to have been first formed in Delaware that same month, and was not even registered to do business in California until January 2021, served a DMCA takedown notice on Twitter, contending that the tweets infringed its copyright in the photos, and followed that notice with a subpoena that it obtained under section 512(h) of the DMCA, seeking to identify CallMeMoneyBags for the claimed purpose of enforcing its copyright.
Twitter objected to the subpoena and, when Bayside persisted, moved to quash on First Amendment grounds, expressing suspicion that Sheth himself was behind the DMCA takedown and subpoena, and insisting that Bayside make a showing that its copyright claim could surmount the anonymous user’s potential fair use defense before the user’s identification was compelled. Bayside cross-moved to compel compliance. The Magistrate Judge to whom the case was assigned initially ordered Twitter to give notice to the anonymous user so that she could, perhaps, retain counsel and articulate her own objections to the subpoena. When the anonymous user failed to take advantage of this opportunity, the Magistrate Judge ordered Twitter to provide the identifying information, reasoning that neither the fair use defense nor the “balancing” stage of the subpoena analysis could be applied properly unless the anonymous accused offender entered the case to provide evidence bearing on her purpose in using the photos or the harms that could befall her if she is identified.
Twitter has sought de novo review, and its position has been supported by EFF and the ACLU, which filed an amicus brief urging reversal of the magistrate judge’s ruling and a grant of the motion to quash. Twitter and its amici argue that the magistrate judge got the fair use analysis wrong, but also that the magistrate judge should have quashed the subpoena based on the balancing stage alone, either because the copyright owner has the burden of proving that its interest in enforcement outweigh any possible First Amendment interests in anonymity (the EFF/ACLU argument), or because it is “self-evident that identification would [result in] public exposure of plaintiff’s identity and the financial and other burdens of defending against a multi-count lawsuit.” (Twitter’s argument) Bayside is defending the decision it secured, also supported by amicus briefs from the Copyright Alliance and from a coalition of photographers' organizations: Their arguments include the contentions that an Internet platform should never be allowed to invoke its users’ possible fair use defenses as a basis for opposing enforcement of a DMCA subpoena, and that, because the defense of fair use supplants possible First Amendment objections to copyright infringement claims, no consideration should be given to the First Amendment right to speak anonymously. They argue that the court should apply the standard of the Sony Music case to decide whether the alleged infringer should be outed.
Public Citizen filed our own amicus brief taking an intermediate position between the two extremes asserted by Bayside and its amici on the one hand, and Twitter and our friends at EFF and the ACLU on the other. We take no position on whether, based on a proper analysis, the subpoena should be enforced. (Many thanks to Phil Malone at Stanford's Juelsgaard Clinic for collaborating on the amicus brief at a difficult time).
Flaws in Big Copyright’s Approach
The fundamental premise of Bayside’s approach this case, supported by its amici, is that copyright law generally, and the fair defense in particular, incorporates First Amendment protections and exhausts the free speech protections that the constitution can provide. Consequently, they argue, there is no need to weigh a user’s First Amendment right to speak anonymously against the copyright holder’s interests in enforcing its rights.
There are several flaws in this argument. Most important, the Dendrite line of cases is not premised on the notion that the enforcement of the various claims over actionable speech violates the First Amendment. Rather, the analysis assumes that valid claims could be litigated to judgment without offending the First Amendment, but demands, based on the First Amendment right to speak anonymously, legal argument and a modicum of evidence supporting the validity of those claims before identifying information is compelled.
For example, just as the infringing use of a copyrighted work to which no sound fair use defense applies is not protected by the First Amendment, so too the publication of false factual assertions about a public figure with knowledge that the accusations are false is not protected by the First Amendment. But the Dendrite analysis nevertheless requires a would-be plaintiff to articulate a viable cause of action and to produce evidence of falsity (and other elements of a prima facie case) because, at the complaint stage of the litigation, when the identity is sought for the purported purpose of pursuing litigation, there has not yet been any showing that the plaintiff does have sound claims directed at the speech. The same may be true in a copyright case.
Moreover, back when we at Public Citizen, along with the ACLU and ACLU of New Jersey, were formulating the test that has become known as the Dendrite test, there were many examples of cases in which companies and others with the resources to hire counsel had used subpoenas to out their critics so that they could be punished outside the litigation, and without any real plan to pursue a suit to judgment. Public Citizen’s amicus brief points to a number of examples of this phenomenon, some of them quite recent.
Another flaw in the Big Copyright approach to this problem is that the statutory limits on copyright law do not exhaust the free speech protections afforded by the First Amendment, particularly at early stages of the litigation when a copyright owner is seeking preliminary relief. A panel of the Eleventh Circuit took considerations of irreparable injury into account in Suntrust Bank v Houghton Mifflin, where it reversed a preliminary injunction against the publication of “The Wind Done Gone” on prior restraint grounds. Although this reasoning did not appear in the later opinion explaining the court’s decision more fully, the en banc court Ninth Circuit in Garcia v. Google similarly relied on First Amendment principles and the rule against prior restraints in vacating a preliminary injunction against the publication on YouTube of a nasty film entitled “Innocence of Muslims,” noting that a “thin copyright claim” “cannot overcome the historical and heavy presumption against such restraints.” Properly applied in an appropriate case, the Dendrite balancing standard allows a court to take equitable considerations into account in deciding a case such as the China hypothetical. This is, as we see it, a necessary aspect of the proper analysis of this sort of case.
Bayside and its amici urge the court to apply the Sony Music standard instead of the Dendrite standard; responding to this argument was one of the most important reasons for Public Citizen to file as amicus curiae. Our brief explains in some detail why the rule in that case, where 17 music companies brought an infringement claim against 40 anonymous Internet users based on their use of peer-to-peer downloading software to participate in the free exchange of copyrighted music recordings, does not provide a good basis on which to decide whether to compel the identification of an anonymous Twitter user who used photos to express his criticism of a wealthy investor. Indeed, as our brief explains, the Sony Music standard is stacked in favor of the copyright holder and is used throughout the country only in cases involving downloading of songs and movies. Only in the Second Circuit are there a handful of district court cases applying Sony Music beyond the downloading context, and we are always on the lookout for good cases to take that on.
Moreover, there is a certain irony in coupling the argument that the judge should disregard the First Amendment altogether, with the contention that the judge should apply the Sony Music standard, because that standard was articulated as a way of accommodating the need to enforce copyright law with what Judge Chin called the very weak First Amendment interests at stake in that case.
The Flaws in the Arguments from Twitter and Its Supporting Amici
But we had a second strong reason for filing our brief – our judgment is that Twitter and our friends at EFF and the ACLU of Northern California have substantially overstated the protections that the balancing prong of the Dendrite standard provides to an anonymous speaker, and that their arguments put the viability of the Dendrite analysis at risk.
When we crafted the Dendrite standard more than twenty years ago, the idea was to provide an extra level of protection for the anonymous speaker, or indeed for a person claiming to have been wronged by online speech, when the equities ran unusually strongly in one direction or another — borrowing from the standard for issuance of a preliminary injunction, when the equities tip decidedly in favor of the proponent or opponent of preliminary injunctive relief. But our assumption all along was that a plaintiff with a solid claim would generally be able to surmount the Dendrite test, obtain identifying information, and proceed with the litigation.
Unfortunately, some of the arguments that Twitter has made in the case, as well as arguments in the EFF / ACLU brief, tend to suggest that the balance should always be struck against identification. That cannot possibly be right— the whole raison d’être for the Dendrite standard is to identify the cases in which it is fair to let the claim of wrongful speech be litigated, weeding out the weak lawsuits where it is inequitable to allow a powerful plaintiff to out its critics or other adversaries. Our brief spells out the ways in which we think Twitter and our civil liberties colleagues have overstated the test, and perhaps overstated the argument for quashing the subpoena. And that overstatement could put at risk the palatability, from a neutral perspective, of the Dendrite test itself. At the oral argument in Dendrite, and in several arguments on this issue since, and in the resulting opinions, judges have made it clear that they accept our proposed test precisely because the plaintiff with a genuine claim can generally prevail.
At the same time, there are serious issues with the ruling of the Magistrate Judge, and with the arguments in support of that ruling posited by Bayside Advisory, with respect to the balancing stage. Indeed, the whole case appears to be quite peculiar, considering that, for aught that appears, Bayside Advisory looks to be a front for whoever it is that is really objecting to the use of the photos, and whoever is spending big money to enable a copyright claim whose outcome cannot possibly provide economic justification for the money being invested. Bayside, while formulating a carefully worded denial of connections with Brian Sheth, has refused to say how it got the copyright in the photos and who is financing its subpoena litigation. When I first heard about the case, I asked some probing questions of Bayside’s lawyer, to which he never responded. Perhaps Bayside will address such issues in its reply brief. Our brief urges the trial judge to demand answers to this sort of question, and suggests various issues that the judge should consider. This may, indeed, turn out to be just the sort of "thin copyright claim” about which the en banc Ninth Circuit expressed concern in deciding Garcia v. Google.
We also urge the judge to consider asking a lawyer in the Northern District’s pro bono panel to represent the interests of CallMeMoneyBags in this case. After all, the Doe may yet have useful information to provide about her specific circumstances and the risks that she faces from being identified. But we take no position on how the judge should rule on the subpoena.
The Usual Suspects Disagree
I have long commended online platforms such as Twitter, Glassdoor and Yelp for their willingness to go the extra mile for their anonymous users, spending their own funds to ensure that Dendrite issues are litigated and that judges focus on the key free speech issues. It is, therefore, not with any joy that I have written a brief disagreeing with Twitter’s position, or with the EFF and ACLU colleagues with whom I have collaborated over the years in promoting the Dendrite approach to this problem. But pitching the test in a way that suggests that the online speaker should never lose is not a good way to persuade a judge to adopt that test. And even though a number of trial judges in the Northern District of California have opted for the balancing approach, trial court rulings are not binding in other cases pending in the same trial court. So it remains up to Judge Chhabria in this case to decide whether the Dendrite test governs here. Our brief urges him to adopt that test, and then to use his best judgment in applying it.