Catching Up on 512 Safe Harbor Cases (and Other Online Copyright Cases) From the Past Year

[Post by Venkat Balasubramani with a comment by Eric; followed by a massive supplement from Eric]

shutterstock_53446765.jpg[Eric’s intro: sometimes, a draft blog post misses the publication window and then comes out painfully late. Venkat drafted the first draft of this recap of 3 Ninth Circuit cases on a timely basis, and I sat on it for months–and in the interim, this post lost some lustre in light of the Viacom v. YouTube ruling. Rather than spike Venkat’s post, I decided to expand it by doing a recap of a bunch of other cases I’d been sitting on for the past year. So this post has gone from an outdated post to a very long and outdated post. Hope you enjoy nonetheless!]

In the space of two weeks in March, the 9th Circuit issued three copyright rulings that address the standards for intermediary liability. The results are mostly good for intermediaries, but they (shockingly) leave some clarity to be desired.

UMG v. Shelter Capital (Veoh) (Mar. 14, 2013) [pdf]: The 9th Circuit previously issued a ruling in this case (covered by Eric here: “UMG v. Shelter Capital: A Cautionary Tale of Rightsowner Overzealousness“). It issues an amended opinion which basically harmonizes the previous ruling with the Second Circuit’s ruling in Viacom. Bruce Boyden at Madisonian has a helpful redline comparing the two rulings.

Here is what I see are the main changes:

1. The court reaffirms that the knowledge requirement in 512(c) requires actual knowledge and not some general knowledge that because things like music videos are uploaded or stored within the system; there must be knowledge of unauthorized content. The court adds language citing to the willful blindness language from Viacom.

2. The court also adds a quote from Viacom’s “red flag” test for 512(c)(1)(A)(ii).

3. The court says that the “right and ability to control” infringing activity under 512(c) requires “substantial influence.” This is different from the vicarious liability standard that looks to whether there is the ability to control (as in Napster). The ability to remove locate and remove information, or filter—all actions Veoh points to in trying to establish its entitlement to the DMCA safe harbor—does not rise to this level. The court gets rid of language that says the right and ability to control must be over activity that relates to infringements the intermediary knows about.

The court in several places talks about the relationship between vicarious liability standards and the DMCA, but does not definitively say anything, except that there will often be substantial overlap between facts relevant to vicarious liability and those relevant to DMCA safe harbor protection.

This seems like mostly good news for intermediaries. In particular, the court seemed to highlight the influence necessary for control, and reiterate that merely being able to locate, remove, and filter content is not the type of control that the “right and ability to control” prong talks about.

The Copyright Alliance says the ruling has silver linings. I think this reading excessively parses the ruling, but check out their view: “Revised UMB v. Voeh Decision Includes Silver Linings for Copyright Plaintiffs.”

Ludvarts v. ATT Mobility, et al. (Mar. 25, 2013) [pdf]: This is an odd case where plaintiffs sought to hold wireless providers liable for allegedly infringing MMS messages transmitted through their systems. The Ludvarts develop and sell “greeting card style messages” that they say can only be shared once. It’s unclear how precisely this limitation is communicated to purchasers or licensees, but it raised an interesting question of whether recipients are even bound by this as a contractual restriction. Anyway, in a move reminiscent of the efforts of early plaintiffs against dial-up service providers, they tried to hold carriers liable for the infringements by their end users, under either vicarious or contributory theories of liability.

Vicarious liability: The court cites to the Napster test for vicarious liability and says that the current architecture of the networks does not confer the “right and ability to supervise” on the networks. The court says that the networks don’t have an obligation to implement any technical features that would allow them to control the content transmitted, and the “metadata system of digital rights management” suggested by the Ludvarts was too speculative.

Contributory liability: Liability under this test requires (1) knowledge of direct infringements and (2) inducement, causation, or material contribution. The court says that the Ludvarts fail to allege knowledge on the part of the networks. They cited to notices that they sent to carriers, but the court says these 150 page long lists of titles are insufficient to impart actual knowledge. In fact, the court says that the DMCA “precludes notices as vague as the notices [in this case].”

Interestingly, the court did not rely on DMCA immunity and merely found that the Ludvarts failed to satisfy Iqbal/Twombly pleading standards for derivative liability.

Columbia Records v. Fung (Mar. 21, 2013): Last but not least, the court issued its opinion in Fung, a case involving torrent sites operated by Gary Fung. (It’s a 59 page behemoth of an opinion that was painful to digest, so I will just summarize the high points.)

Inducement liability: The court says that summary judgment was appropriately granted on the issue of inducement liability. Grokster’s inducement theory is not limited to ‘devices’. According to the court, there was more than enough evidence of Fung’s statements that were “designed to stimulate others to commit [copyright] violations.” These included:

responding personally to queries for assistance in: uploading torrent files corresponding to obviously copyright material, finding particular copyrighted movies and television shows, getting pirated material to play properly, and burning the infringing content onto DVDs for playback on televisions.

The court adds some equivocal language about how a loose standard of causation can result in unlimited liability and chill innovation and online commerce, but that’s not much help to Fung who at this stage is contesting only the determination of liability and injunctive relief.

DMCA safe harbors: First off, the court says that DMCA protection and vicarious liability are not mutually exclusive (citing to the 9th Circuit’s amended opinion in Veoh, mentioned above):

we are not clairvoyant enough to be sure that there are no instances in which a defendant otherwise liable for contributory copyright infringement could meet the prerequisites for one or more of the DMCA safe harbors. We therefore think it best to conduct the two inquiries independently—although, as will appear, aspects of inducing behavior that give rise to liability are relevant to the operation of some of the DMCA safe harbors and can, in some circumstances, preclude their application.


512(a) safe harbor: This safe harbor is available to service providers who act as “conduits”. The court says that Fung’s trackers (that manage a swarm of connections related to a particular file) do more because they “select which users will communicate with each other.”

512(c) safe harbor: This was the safe harbor at issue in the Veoh case. The court says that the district court wrongly rejected this safe harbor on the basis that the infringing material does not “actually reside on Fung’s servers.” The 9th Circuit says this safe harbor is slightly broader, and covers “infringing activities that use . . . material stored on the system or network.” Nevertheless, the court says that Fung cannot avail himself to the 512(c) safe harbor because he had knowledge of infringing activity. Fung argued (reasonably) that the DMCA notices sent by Columbia were non-compliant and should have been disregarded. The court says these notices don’t matter, because Fung had “red flag” knowledge of a “broad range of infringing activity.” The court only says that Fung helped users with respect to material that was “sufficiently current and well known” that it would be obvious to a reasonable person that these were not licensed and therefore infringing.

Fung loses protection due to financial benefit: The court also says that Fung is ineligible for protection because he (1) receives a financial benefit directly attributable to infringing activities and (2) he had the right and ability to control the infringing activity. The court canvasses the financial benefit cases and says that Fung’s income stream “derived from advertising” meets this prong:

Fung promoted advertising by pointing to infringing activity; obtained advertising revenue that depended on the number of visitors to his sites; attracted primarily visitors who were seeking to engage in infringing activity, as that is mostly that occurred on his sites; and encouraged that infringing activity.

The court also says that he had the “right and ability to control,” citing to his organization of torrent files, removal of bad files, and assistance provided to users.

512(d) safe harbor: Finally, the court summarily says that Fung is not entitled to this safe harbor because he was aware of facts or circumstances from which infringing activity was apparent and received a financial benefit from activity that he had the right and ability to control.


The 9th Circuit’s Fung opinion is clunky. Interestingly, this case was heard by the exact same panel as the Veoh panel, which just goes to show that you’re not guaranteed a consistent ruling from the 9th Circuit, even if you have the same panel.

Fung was teed up as a case that would resolve whether DMCA safe harbors precluded derivative liability, or at least better explain how the two fit together, but the court punts on this issue. (Google weighed in as amicus for a few reasons, including to argue this point.) While the court makes statements that taking steps that help you comply with safe harbor provisions should not be used against you, we’re still unclear on what derivative liability survives DMCA safe harbor. Given the court’s tortured discussion of several DMCA issues, perhaps it would have been preferable for the court to draw a line and categorically say that if you are liable under a Grokster-inducement liability theory, then DMCA safe harbors are not available to you.

The most painful part of the Fung opinion is its discussion (pp 46-53) about the “red flag” trigger for knowledge, as well as the financial benefit and right and ability to control. I found it difficult to meaningfully differentiate between the type of knowledge and financial benefit that Fung had from what Veoh had. In particular, both defendants generate financial benefit from advertising, and the basis for concluding Fung had knowledge of underlying infringements were fairly slim. The court seems to say that because they are “well known titles” or “top movies,” they are obviously infringing. The court rests its conclusion that there was “red flag” knowledge on precisely the same stuff that UMG argued should result in knowledge in Veoh–the works were popular! they were current! of course they were unauthorized! We are all aware of well publicized instances where material that was thought to be infringing turned out to not be infringing and vice versa. I would say this is a good candidate for en banc review for a variety of reasons, but the muddled discussion about red flag and financial benefit alone makes me hope that the court take the opportunity to clear things up. [Added: the 9th Circuit denied en banc review.]

It would be easier to reconcile the two if Fung distributed a Napster-like product and the ruling can be based on some species of P2P exceptionalism, but this isn’t necessarily the case.

Interestingly, Eric’s post-Viacom reaction was that it’s a bummer for UGC sites and intermediaries. The 9th Circuit’s changes to the UMG opinion followed by this ruling tend to point in the direction that he may be right.

Previous posts:

UMG v. Shelter Capital: A Cautionary Tale of Rightsowner Overzealousness

Second Circuit Ruling in Viacom v. YouTube Is a Bummer for Google and the UGC Community


Eric’s Comment: I stand even more firmly behind my conclusion that Section 512 is a misdesigned safe harbor. My rule of thumbs are that P2P defendants will be denied 512 defenses and plaintiffs who didn’t send 512(c)(3) takedown notices will lose (eventually), but still I really can’t confidently predict what the Ninth Circuit–or any other court, for that matter–will do in any particular case. I can predict with confidence that 512 defendants will spend lots of money (hello, UMG, Veoh went bankrupt, why are you still kicking the corpse?) to assert the safe harbor if contested by a well-funded plaintiff, which is my definition of a low-efficacy safe harbor.


Eric’s Supplement

BONUS: In addition to sitting on this post for months, I’ve had a lot of online copyright cases languishing in the queue. Here’s a recap of other online copyright cases we haven’t covered on the blog over the past year:

Capitol Records, Inc. v. MP3tunes, LLC, 2013 WL 1987225 (S.D.N.Y. May 14, 2013). Prior blog post.

Terrible ruling. The parties ask the court to reconsider its 2011 ruling in light of the Second Circuit’s Viacom v. YouTube ruling. It’s a good sign that the Second Circuit clouded the legal standard rather than clarifying it when both parties think the ruling helps them. Thanks so much, Second Circuit!

Even worse, the court reverses the summary judgment it granted to MP3Tunes because Capitol argued “willful blindness.” What exactly is willful blindness? I don’t know, this judge doesn’t know, and the Second Circuit doesn’t know, though they will get more chances to restate what they meant. Here’s the evidence this court says trumps its already-granted summary judgment:

an email received by MP3tunes in April 2007 gives a specific blog title and states, “[a]lthough I don’t like ratting myself out, everything I post is in clear violation of the DMCA …. please remove any MP3s that are linked to that site.”…Another email from November 2007 states, “if you search for ‘the clash I fought the law’… you will get 5 results … 2 of which point to the website[.] This website blatantly acknowledges that it contains infringing MP3’s.”…In a third email, an MP3tunes employee acknowledges that while “it’s not clear if [content from a user’s site] is all copyright [sic] material … it probably is though.”

Good luck finding any service provider that doesn’t have similar emails in its archives. And as we feared from the Viacom v. YouTube ruling, it appears that emails sent by non-copyright owners–that by definition don’t comply with 512(c)(3) and therefore putatively are outside the 512 statutory framework–now appear fully relevant to the willful blindness inquiry, which also isn’t referenced anywhere in the statute. How did we reach this statutory interpretation again???

The court similarly concludes the Second Circuit broadened the “red flags” standard, as that too leads to a trump of summary judgment for MP3Tunes. The court says:

Since something less than a formal takedown notice may now establish red flag knowledge and EMI offers communications acknowledging likely infringement, the issue of Defendants’ red flag knowledge cannot be resolved on summary judgment. This Court reaches this conclusion reluctantly, given MP3tunes’ salutary practice of sending instructions regarding DMCA-compliant takedown notices to third parties reporting possible infringement and the DMCA’s disavowal of any duty on the part of service providers to monitor user content.

You can practically hear the judge muttering swear words at the Second Circuit under his breath. But isn’t this result terrible? Does this suggest all “red flags” issues are jury questions?

The judge sidesteps questions about the interplay between inducement and the 512 safe harbors, saying that Capitol never provided enough evidence of inducement to make a tenable claim. I wouldn’t be surprised to see Capitol take that issue up on appeal.

The court reiterates that 512 doesn’t apply to MP3Tunes’ practice of automatically gathering some album cover art from There’s yet more interesting stuff in the opinion if you want to relive the horror.

Football Ass’n Premier League Ltd. v. YouTube, Inc., 2013 WL 2096411 (S.D.N.Y. May 15, 2013).

No class adjudication of the issues in the parallel lawsuit to Viacom v. YouTube. I still think copyright lawsuits are categorically poor candidates for class consolidation.

Associated Press v. Meltwater U.S. Holdings, Inc., 2013 WL 1153979 (S.D.N.Y. March 21, 2013).

This is a pretty significant opinion, and it deserves much more careful treatment than I’m going to give it here. Meltwater provides “an Internet media monitoring service” for public relations professionals, i.e., it allows clients to set keyword alerts and find documents containing those keywords. Meltwater builds its database by scraping the web and delivering to its clients excerpts of documents matching the keywords.

Because it scrape the web, it picks up content from lots of copyright owners, including Associated Press. In this respect, Meltwater functions like any other search engine; the main difference is that Meltwater users subscribe to the database, as opposed to Google’s ad-supported main search index or its (typically?) ad-free Google News database. It’s probably also relevant that AP offers a licensing program for article excerpts, and three of Meltwater’s competitors have paid up. The court also cites some evidence that AP and Meltwater also directly compete with each other in some circumstances.

The court denied Meltwater’s fair use defense. The court says there’s no transformative use:

Meltwater copies AP content in order to make money directly from the undiluted use of the copyrighted material; this is the central feature of its business model and not an incidental consequence of the use to which it puts the copyrighted material.

The court also says the public interest weighs against Meltwater:

Permitting Meltwater to take the fruit of AP’s labor for its own profit, without compensating AP, injures AP’s ability to perform this essential function of democracy….the public interest in the existence of such commercial enterprise does not outweigh the strong public interest in the enforcement of the copyright laws or justify allowing Meltwater to free ride on the costly news gathering and coverage work performed by other organizations. Moreover, permitting Meltwater to avoid paying licensing fees gives it an unwarranted advantage over its competitors who do pay licensing fees.

Because Meltwater has a low clickthrough rater (as low as 0.08%), the court says Meltwater isn’t like a general-purpose search engine: “Instead of driving subscribers to third-party websites, Meltwater News acts as a substitute for news sites operated or licensed by AP.” The court distinguished Perfect 10 v. Amazon and Kelly v. ArribaSoft because (1) they involved photos, not text, and (2) Meltwater is a subscription search engine, not open to all. The court sarcastically says “use of an algorithm to crawl over and scrape content from the Internet is surely not enough to qualify as a search engine engaged in transformative work.”

Regarding the amount and substantially of the portion taken, the court says this weighs heavily against Meltwater because “Meltwater took between 4.5% and 61% of the Registered Articles. It automatically took the lede from every AP story….search engines regularly display briefer segments of news articles.”

Given that AP had already procured licenses from Meltwater’s competitors, it’s not surprising that the market effect factor weighed against Meltwater as well. (As usual, the court doesn’t address the circularity/bootstrapping in that conclusion).

The court rejects Meltwater’s implied license defense. In particular, the court rejects the argument that AP should have required licensees to use robot exclusion headers (robots.txt), and its failure to do so created the implied license. I don’t recall such a thorough treatment of the subject before. In fact, I did a quick search in Westlaw’s ALLCASES database and I found only 6 other cases referencing robots.txt (including Parker v. Yahoo, Field v. Google, eBay v. Bidder’s Edge and the ArribaSoft case). Distinguishing the Parker and Field cases, the court says “Meltwater does not provide its subscribers with access to cached pages, reserves the right to disregard certain robots.txt instructions, and has not suggested that it will remove content from its system at the request of the copyright owner.”

After Meltwater lost this ruling on summary judgment, the parties entered into a settlement eliminating all of the other claims in the case (other than the direct copyright infringement claim) and are wrangling over the injunction’s scope, presumably in preparation for an appeal.

There’s yet more interesting discussion in this case. It could very well make my year-end list of cases. If you’re in the scraping business, this opinion is a must-read.

Datatech Enterprises LLC v. FF Magnat Ltd., 2013 WL 1007360 (N.D. Cal. March 13, 2013).

This is a lawsuit against a Hong Kong cloud-based locker who is trying to invoke a 512 defense. The court says:

Datatech’s evidence shows that when Oron learned that particular users were engaged in extensive repeat infringement of U.S. copyrights—for example, a single user who uploaded 1,600 separate copies of infringing work—Oron regularly declined to ban them despite requests from copyright holders….At the hearing, Oron’s counsel represented that Oron attempted to ban the particular user who had uploaded 1,600 files, but nothing in the record reflects any such attempts. The Court accordingly reaffirms its conclusion that Datatech has shown a likelihood of overcoming any safe harbor defense, making unnecessary discussion of Datatech’s promising alternative argument–the DMCA’s requirement that the service provider not “receive a financial benefit directly attributable to the infringing activity.”

Perfect 10, Inc. v. Giganews, Inc., 2013 WL 2109963 (C.D. Cal. March 8, 2013).

I mentioned this post in my Section 230 catch-up post. You already know Perfect 10; the defendant sells commercial access to USENET.

I think this case has among the most sophisticated judicial discussion of “volition” in direct copyright infringement that I can recall seeing. Judge Matz writes:

by imposing a requirement of “volitional conduct” Netcom merely focused the inquiry on whether the defendant directly caused the infringement to take place…Plaintiff has not alleged that Defendants were the direct cause of, or actively engaged in, direct infringement….The Court disagrees with the application of the volitional act requirement in Usenet, MegaUpload, Capitol Records, and Playboy Enterprises. By focusing on the defendant’s awareness or state of mind–rather than on who actually caused the infringement–these cases effectively hold defendants liable for copyright infringement committed by third parties without requiring a full assessment of the additional elements of secondary copyright infringement claims.

As a result, Giganews lacked the volition to commit direct infringement.

As usual, Perfect 10 didn’t send a typically-formatted 512(c)(3) takedown notice. Still, in the context of a motion to dismiss, the court says Perfect 10’s notice was sufficient to constitute knowledge for contributory infringement purposes (I’d have to go research the status of a 512 defense, not mentioned in this ruling):

Plaintiff has alleged that it sent Giganews a notice that identified hundreds of Perfect 10 copyrighted images that Plaintiff had found using Defendants’ service, some of which displayed a copyright notice. Plaintiff also alleges that Giganews has a search function which Giganews could have used to search for the infringing content based on the image identifiers Plaintiff provided in its notice. The Court finds these allegations sufficient to support Plaintiff’s claim that Giganews had “knowledge” of the content infringing Plaintiff’s copyrights.

The opinion also has an interesting discussion about the direct financial interest prong of vicarious infringement. Giganews charged a flat monthly subscription price, and ordinarily that should be enough to avoid direct financial interest in infringement because the price doesn’t vary based on the amount of infringement. The court says:

Plaintiff alleges that USENET is now used primarily for pirated content and that the availability of the pirated content is what attracts most of Defendants’ monthly customers…In other words, Defendants allegedly charge a fixed fee for a service that mainly exists to provide access to the infringing material on USENET. At the motion to dismiss stage, the Court must accept these allegations as true.

In contrast, Perfect 10’s trademark claims (direct, contributory and dilution) fail because Giganews sells USENET subscriptions, not the individual works, so Giganews didn’t make a “use” of Perfect 10’s trademarks.

David v. CBS Interactive Inc., 2:11-cv-09437-DSF-JC (C.D. Cal. Feb. 19, 2013). Prior blog post.

This is a troublesome “tertiary liability” case. The plaintiff seeks to hold for users’ download of P2P software which was allegedly used to infringe. In other words, the web host (the tertiary infringer) would be liable for providing software technology (the secondary infringer) that users misuse (the primary infringer). The court previously rejected the secondary infringement claims against the web host, but nevertheless said an inducement claim could be tenable. In the February ruling, the court refuses a preliminary injunction because it’s unlikely the plaintiffs can show the defendants are liable for any continuing infringements:

While there might be some evidence of past inducement of copyright infringement, there is no evidence of any ongoing distribution of any file sharing software “with the object of promoting its use to infringe copyright, as shown by clear expression or other affirmative steps taken to foster infringement.”

The court continues:

Plaintiffs’ only solid evidence of possible inducement comes from reviews that were published a decade ago…The other articles cited by Plaintiffs merely discuss P2P issues, including legitimate distribution through P2P, and the various technological and legal issues that have emerged with the technologies….The Court has no reason to believe that Defendants will purposefully encourage copyright infringement now or in the foreseeable future….Most of the articles cited by Plaintiffs are straightforward, legitimate news articles that do not in any way encourage or induce copyright infringement. This suggests that Plaintiffs’ goal goes far beyond stopping actual infringement by Defendants and extends instead to silencing public discussion of P2P technologies.

It would be great if this meant the judge is beginning to realize that the earlier inducement ruling is problematic, especially when there’s no other secondary infringement. I’m not sure the judge will reach that epiphany, unfortunately.

Amaretto Ranch Breedables v. Ozimals Inc., 2012 WL 5389897 (N.D. Cal. November 5, 2012). I blogged several prior developments in this case.

For anyone who wants to know why I still love Internet Law after all these years, take a look at this case. It’s a dispute between providers of virtual bunnies and virtual horses in the virtual world Second Life, and it’s soooo interesting to me! This particular ruling is pretty straightforward doctrinally; it eviscerates the bunnies’ copyright claim against the horses. The parties are still wrangling over other issues, though, undoubtedly to the massive financial detriment of all involved.

Does 1-4 v. Arnett, 2012 WL 3150934 (C.D. Cal. August 1, 2012).

A super-interesting dispute involving copyright as a right to forget. To remove a photo from the Internet, Arnett (the photo subject) claimed copyright and sent a 512(c)(3) takedown notice followed by a 512(h) unmasking subpoena request. The host took the photo down, and Arnett withdrew the 512(h) subpoena, but the Does pressed a declaratory judgment action, as well as a 512(f) claim. Sadly, we don’t know how the legal doctrinal story ends because the parties settled in April.

American Broadcasting Company v. Aereo, 1:12-cv-01543-AJN (SDNY July 11, 2012), affirmed by the Second Circuit in April.

This case is so old, I will only address it briefly. I’ve always been troubled by the Cablevision precedent because it rewards form over substance–Cablevision’s DVR is legally OK only because it was designed with socially wasteful resource redundancy. Aereo takes that precedent and extends it to even more ludicrous extremes, where Aereo deploys thousands of little antennae to accomplish what it can’t legally do using a single antennae. The problem isn’t the courts–they are reaching realpolitik outcomes–but the technology-dependent overengineering of copyright law, done to protect business models as implemented at fixed points of time with embedded technological assumptions. I doubt we’ll rid copyright law of that defect, so instead we get these tortured interpretations of the law and situations where the lawyer’s irrationality trumps engineering common sense. Sigh. James Grimmelmann also explores this issue better than I can.

[Photo credit: 3d stamp delayed // ShutterStock]