In Its “Innocence of Muslims” Ruling, the Ninth Circuit is Guilty of Judicial Activism–Garcia v. Google

In a ruling that sent shockwaves through the internet community, the Ninth Circuit, with one judge dissenting, found that an injunction should have been granted against YouTube, requiring removal of the “Innocence of Muslims” film.

Does Garcia have a copyright interest? The crux of the dispute—between Chief Judge Kozinski and the dissenting Judge Smith—is whether Cindy Garcia has a copyright interest in her performance.  Judge Kozinski says remarkably that a performance that’s part of a larger film may itself constitute an independently copyrightable “work”.

Citing to authorities on acting, Judge Kozinski says that even if the director controlled all aspects of her performance, the actor still imbues her performance with sufficient creativity because she must “live [the] part inwardly, and then give to [her] experience an external embodiment.” Thus, Garcia owns the copyright in the footage that constitutes the manifestation of her performance. Having found a protectable copyright interest, the court goes on to consider and shoot down the two possible bases for the director to nevertheless use her “work”: (1) Garcia was not an employee so the director cannot claim that the footage was a “work for hire,” and (2) the director’s deception and the fundamentally different nature of the end product undermines any possible implied license argument.

The dissent says that an acting performance is not a “work” under the Copyright Act; the film and even individual segments do fall into this category. Garcia doesn’t get past the first hurdle of the copyright analysis. Even assuming the performance is a “work,” the author of that work is the creative engine behind it, not people who merely implement—as Garcia acknowledged she’d done—the ideas of others.  (The dissent also says that an actor’s performance is not fixed—the edited, final version, which the director is responsible for, is.)

The dissent also rips into the majority’s work-for-hire analysis, noting that the key element is the degree of control exercised by the putative employee, and here it was undisputed that Garcia exercised little or no control over the production. Indeed, she was duped.

Has she showed irreparable harm? The court says that she has shown irreparable harm because of the threats to her life.  Despite paying lip service to the fact that the irreparable harm must be sufficiently connected to her cause of action (“she needs to show that the harm she alleges is causally related to the infringement of her copyright”), the court glosses over this issue and does not point to anything in the record as possibly justifying irreparable harm from a copyright owner’s standpoint.

The dissent tees off on this issue as well, arguing that not only is there no irreparable harm from a copyright standpoint, the cat is out of the bag: the film has been widely disseminated so it’s unclear how restricting further dissemination can remedy any harm to her. The dissent also points out that the balance of equities and the public interest—namely the First Amendment—strongly favors leaving the video up.


Yowza, this is a blockbuster ruling. It sent shockwaves through the cyberlaw community, so when the inevitable petition for rehearing is filed, expect to see a long list of who’s who among internet companies as potential amici. (See Prof. James Grimmelmann’s twitter feed for a strong reaction and trenchant critique.) Mike Masnick also does a great of job critiquing the decision at TechDirt, and points out that the decision has the distinction of uniting both the MPAA and Google, which is no small feat. Check out the EFF blog post here.

Interesting procedural note that is also somewhat of a shocker: as noted in this piece from Eriq Gardner at THR, Google received an order in advance of the ruling requiring it to take down the video and not say anything about this. It fought that order and lost the initial round. On a related note, it’s surprising that the court did not just remand for additional factual development or allow the district court to revisit the request for an injunction in light of the Ninth Circuit’s guidance on the legal issues. Reversal and instructing the district court to issue the injunction (and issuing an order in the interim requiring Google to remove all available copies along with a gag order) is over the top to say the least!

As to the copyright question, the conclusion that a performance that’s admittedly part of a larger film can itself constitute a copyrightable work seems out of whack. (A possible boon for those seeking damages for infringements of film.)  Another concern is whether the resolution of the ownership issue comports with the business reality of everyday negotiations between studios and actors. Money is changing hands, and do actors really expect to have some say in the ultimate exploitation of footage they appear in? The majority’s response to this concern is that studios have an available method of protecting themselves: they can obtain contractual releases beforehand. This still seems like a dangerous state of affairs, where a studio is required to obtain a release from every single person who appears in footage or risk a copyright claim and worst yet an injunction requiring the entire film to be put on hold. Yikes! This is an even greater concern when it comes to smaller, less sophisticated content creators, whom the majority derisively refers to as “any schmuck with a video camera.” As to the question of whether someone appearing in footage possibly has a copyright interest in the footage, that seems somewhat less controversial, although still a remarkable conclusion given the facts in this case.

Speaking of which, the court’s conclusion on the lack of an implied license was also surprising. The court says that if you appear in footage for pay, in the absence of a written agreement, the license acquired by the director or producer only extends to use of the footage in the manner you envision.  Ouch. This amounts to a moral right to veto uses of footage that are not to your liking. Presumably, this rule would not apply to directors who have obtained written releases, but the Sacha Baron Cohen’s of the world should be concerned about this. (See “Borat Tricked Me“.)

It’s interesting that Google chose to fight this ruling. Intermediaries often fight to keep content up, but in the copyright context, this is part of an emerging trend to battle wrongful takedowns. (WordPress suing over wrongful takedowns is another recent example.) Is Google now liable for not having taken down the video? [See Eric’s comments below for discussion about the Section 230 issues.] Maybe what’s driving the majority’s decision is cynicism over Google’s decision to fight the takedown request. It receives and routinely complies with thousands of takedown requests—what were its reasons for fighting this particular one, and how exactly does it decide which takedown orders deserve Google’s First Amendment sympathies and the full force of its legal department? Interestingly, Google did not raise fair use as an argument. Not sure what to make of this, given its obvious applicability to Google’s publication of Garcia’s portion of the video. Indeed, it would seem that Google should be able to publish the video sans Ms. Garcia’s performance and perhaps even use the performance or snippets of it in the context of an overall fair use discussion.

Finally, what to make of the free speech implications of this ruling? It’s certainly surprising to see a First Amendment stalwart such as Kozinski casually refer to the Garcia’s dialogue as “fighting words,” and to place such heavy reliance on the threats Garcia received. Compare the tone of this ruling with Kozinski’s opinions in cases such as City of Norse, where he take a much firmer and almost contemptuous stance on the ability of those offended to impose restrictions on speech. Given the well-documented efforts of censorship that use copyright as a proxy, it’s unfortunate Chief Judge Kozinski did not make the connection between the two. I wouldn’t go so far as to say that anyone who issues a DMCA takedown request is a censorious thug, but here, the connection between the actions of those who disagreed with the message and the resulting pressure on Garcia to try to get the video pulled were well documented. This starts to look a whole like a variation on the heckler’s veto. It’s certainly not very speech friendly to take an expansive view of copyright in connection with takedown requests that are prompted by threats of violence. What’s next? Someone who issues a widely ridiculed religious pronouncement starts to issue DMCA takedown notices directed at articles critiquing and poking fun at the pronouncement?

One thing is for sure, that people depicted in films may have a copyright interest allowing them to issue takedowns is an issue of significant interest to big media companies and intermediaries alike. That anyone appearing in a movie or a YouTube video without a written release can now make a colorable threat to have the video pulled via a DMCA notice and an injunction is a big deal. (Sidenote: possible beneficiaries of this ruling potentially involve aggrieved revenge-porn victims and those who are trying to suppress sex tapes.)

Case citation: Garcia v. Google, Inc., 2014 WL 747399 (9th Cir. February 26, 2014)

Eric’s Comments:

[Note: Venkat and I wrote our portions independently, so you’ll see some similarities–and differences–in our assessments.]

This opinion sucks rotten eggs. It is so terrible that there’s simply no point trying to make sense of it. I’m fascinated by well-meaning folks who have tried to treat it as a “serious” statement of the law. It’s not. It’s a one-off hack of the law, and treating it as anything more will take you towards insanity.

Why Things Went Wrong

Why did this opinion go so wrong?

1) Bad facts #1: At the core of this opinion is the movie producer’s lie to Garcia that caused harm that Garcia could have never foreseen. The law punishes falsity extensively, but occasionally we run into situations where our legal tools don’t provide a satisfying response to a deception. In those cases, judges often find elasticity in existing legal doctrines to combat what they view as socially unacceptable behavior. That rarely turns out well for the coherence of legal doctrines.

In this case, copyright law is being asked to cure the movie producer’s lie. Copyright law wasn’t built to solve this problem, so the only way to achieve that result is by breaking copyright law. It’s the same analytical error made by the folks advocating for expansive copyright interpretations to fix revenge porn. Rebecca Tushnet does a much-needed takedown of those arguments in her forthcoming article, How Many Wrongs Make a Copyright?

2) Bad facts #2: The real bad actor is the movie producer who lied to Garcia and then gave her a false edit. However, the movie producer didn’t show up in court. Instead, Google stood in the movie producer’s shoes trying to defend his behavior. But Google wasn’t a party to their conversations or their transactions, and Google’s interest isn’t in defending the movie but in avoiding liability for third party content. So having Google act as a legal proxy for the movie producer is a terrible way to set precedent for the entertainment industry.

3) Bad facts #3: This case came to the Ninth Circuit in an unfortunate procedural posture: seeking a preliminary injunction, which the lower court denied in a perhaps-too-short opinion. The standards for preliminary injunction allow judges to make assumptions about the facts and to overlook obvious holes in the legal theories.

4) Judicial activism. This opinion is a textbook example of judicial activism, i.e., coming up with newly manufactured legal doctrines to find a remedy for a victim. Sometimes, judicial activism leads to brilliant new legal theories that improve the law. Most of the time, it just horks the law and breeds disrespect for the integrity of the legal system.

5) Judge Kozinski #1. Judge Kozinski is a Big Personality. It takes a lot of gumption and intellectual rigor to stand up to him. Judge Smith gave a spirited dissent, but it wasn’t precise enough (and, IMO, rigorous enough) to curb Judge Kozinski’s intellectual excesses.

6) Judge Kozinski #2: My speculation is that Judge Kozinski has schizophrenic attitudes about Internet intermediary liability. Recall in how he said the Internet had outgrown its swaddling because there was a new class of Internet incumbents, and recall in Perfect 10 v. Visa how he was perfectly willing to throw Visa under the bus. I infer from these and other datapoints that Judge Kozinski generally supports reduced liability for Internet intermediaries except when he doesn’t; in other words, he’s fine that intermediaries typically aren’t liable for user content, but in edge cases, he’s willing to ignore those doctrines to achieve what he thinks is a more just result. Because of his idiosyncracies, I imagine this opinion would have looked very differently if Judge Kozinski had not been on this panel. Having Kozinski on the panel was an unlucky draw for Google–and unfortunately for the rest of the Internet and everyone else harmed by this activist opinion.

Errors in the Opinion

It would take me days to make a complete list of what’s wrong with this opinion, Let me just highlight five:

1) Too tricky by half. Judge Kozinski seemingly thought he found a legal loophole to help Garcia without hurting others. He could find a copyright remedy for Garcia by saying that she had a copyrightable work, but he could limit any negative impact of that conclusion by saying almost all video producers in the future will be protected by an implied license (or an express contract) and thus this is the “rare case” where the actor wins. It’s impressively clever legal reasoning if it works, but it’s too tricky by half. Garcia gets her win, but at an enormous social cost.

By shifting the legal inquiry from copyrightability to the terms of licenses, we move from a potential bright-line rule to a standard that will require custom fact adjudication. The practical effect from this opinion (if it stands, which it shouldn’t) is that actors–and perhaps anyone else captured on video–can routinely assert a copyright claim and then impose costs on the movie producer to defend on express or implied license grounds. When there’s a written contract, we expect the debate will go quickly. But in the many cases where there isn’t an written contract–ranging from “candid camera” videos to Sacha Baron Cohen movies to pornography–this opinion means the actors can claim copyright and set up a fact dispute that will be hard to predict and expensive to resolve. In turn, this gives actors a powerful tool to suppress content they don’t like or to play hold-up games for money, and it means intermediaries like YouTube will err on the side of actors’ assertions rather than do the fact investigation. So this ruling should terrify the entertainment industry as well as the Internet industry, and it should increase transaction and negotiation costs for little new incremental social value.

2) The silent role of 47 USC 230. The opinion doesn’t mention Section 230 at all, even though the statute casts a long shadow on the discussion. If Garcia had claimed anything other than federal copyright, her claim against YouTube would have been preempted by Section 230. Defamation, intentional infliction of emotional distress, invasion of privacy, breach of contract, fraud, state copyright, state publicity rights–any of those claims might be viable against the movie producer, but all of them are preempted against Google. The only hope Garcia has to force YouTube to act is copyright law. But by not discussing the interactions between copyright and Section 230, this opinion doesn’t pay any attention to Section 230’s importance, the reasons why it works as well as it does, or how plaintiffs are chomping at the bit to use copyright to work around it (e.g., Medical Justice’s depraved scheme, the Goren case, Scott v. WorldStarHipHop and many others). This case just gives further encouragement to that trend, and thus it poses a huge threat to Section 230’s integrity.

3) The fixation problem. The court says actor’s performance of a script can have copyrightable expression. I’m OK with that in concept. An actor can make a dramatic reading of a script where the actor infuses the reading with his or her own expression, just like cover songs can be radically different even though they are starting with the same music and lyrics. However, in most cases, the actors can’t claim a federal copyright in their performance because of the fixation requirement. In order for Garcia to have a federal copyright, she must “fix” her performance by recording it herself (which she didn’t do) or have the movie producer fix her performance “on her behalf” (also clearly not the case). So there is simply no way for Garcia to claim a federal copyright interest because she failed to satisfy one of the formalities. This will be true of almost all actors who are depicted in video produced by someone else.

(I’m sidestepping some complexities here, such as whether Garcia actually added sufficient independent creativity to her expression, whether Garcia could claim state copyright law for an unfixed work of authorship, and if Garcia can validly assert her copyright interest in her performance when it’s a derivative work of the underlying copyrighted script).

Judge Kozinski knows about copyright law’s fixation requirement. He brushed it aside in FN 4, saying:

Neither party raised the issue of whether the author of a dramatic performance must personally fix his work in a tangible medium. Because the question is not properly before us, we do not decide it. The parties are free to raise it in the district court on remand.

That helps his results-driven decision, but it’s maddening as a caveat to a potentially crucial precedent. Given that fixation is likely to be a dispositive issue, this opinion is built on a foundation that we know will collapse when the fixation issue is examined. As a result, this opinion is worthless as precedent because it deliberately bypassed a dispositive issue. If there is an en banc hearing and the fixation issue is litigated, I’m 99% confident the fixation issue will determine the outcome (I say only 99% because, after all, this is the Ninth Circuit we’re talking about).

4) The secret order. What is it with government entities issuing secret orders to Internet companies? Is that some fashionable new trend? I must not be very fashionable, because I thought I lived in a country where we didn’t do such things. Didn’t the Ninth Circuit learn anything from the NSA scandals?

In this case, the secret order is especially baffling because it was completely pointless. The video went viral a couple of years ago, so there is little risk that there will be a new rush to copy and repost the video immediately after a court removal order. Indeed, it’s usually impossible to erase content from the Internet, especially viral content. Even if YouTube successfully wipes the video off the network, it will be freely available forever on other services, including those outside the United States. So having an appellate court issue a secret order to remove Internet content sets a terrible precedent and will have zero effect on solving Garcia’s problem.

5) Erasing history. Even if the video never should have been published, it’s now an irreplaceable part of history. We cannot fully appreciate the Benghazi attacks or other anti-American attacks that may be attributable to the video, or the fatwa against Garcia, or even this opinion without seeing the video itself. This opinion is spiritually related to the efforts towards a “right to forget.” That entire movement is an assault on our notions of historical developments and how we learn from history.


This is a rare opinion that unites the movie studios and Google in outrage. The opinion screws over both industries, and (as a “bonus”) the public as well. I can’t wait for the opinion to be scrapped, as it so richly deserves.