Secondary Liability and Safe Harbors
Moderator: Andrew Bridges, Fenwick & West LLP
Shira Perlmutter, USPTO
§512 was intensively negotiated and complex; intended to be a very careful balancing act, so carefully balanced that the participants would’ve told you that every small twig was important to the entire edifice. Protection against monetary remedies for those ISPs who acted responsibly. Initially controversial, but did begin to function as alternative to litigation, and over the years exported widely as a model, including to the European Union, Japan, and China, and others primarily in Asia. 15 years later, strains inevitably began to appear (RT: which is of course why they called it the Digital Millennium Copyright Act) as flaws from real life application and technological evolution occurred. Target in 1998 was websites hosting infringing content, but now we have P2P, cloud services/cyberlockers, and streaming services (some licensed and others not). The four covered activities still broadly cover ISP activities—transmitting, hosting, caching, and search. But the big question is whether the law is still fit for purpose.
Many issues have been or still are playing out in the courts. Among the questions: who qualifies as ISP? How do safe harbors relate to inducement liability after Grokster? What’s the meaning of red flag knowledge? Settlement of Viacom v. Google makes that more elusive. What kind of financial benefit is relevant? What are appropriate repeat infringer policies? What’s the scope of the subpoena process laid out in §512?
What isn’t covered today and needs attention? What needs fixing in § 512? Those are the basic debates. Some things people complain about: P2P, though may be lessening in importance. Addressed in voluntary copyright alert system. Other area pointed to (RT: by whom?): foreign sites dedicated to infringement, SOPA/PIPA, spectacular failure. Also some voluntary initiatives using the “follow the money” approach with advertisers and payment providers.
Operation of notice and takedown: 3 main issues. (1) Inefficiencies in the system, including huge volume of notices, and phenomenon of rapid reappearance of content taken down. (2) Misuse of notices for fair use, political campaigns, copyright trolls. (3) Special difficulties for individuals and small and medium sized enterprises, whether rightholders or ISPs.
Initiatives on Capitol Hill: copyright review sparked by Pallante’s speech. Dep’t of Commerce Internet Policy Task Force—Green Paper. All sectors perceived problems with the notice and takedown system, though some entities within each sector were ok. Some entities wanted to see legislative change, but most of them strongly opposed reopening the statute. Given the nature of the problems identified and given problems of moving copyright legislation forward, especially in the online enforcement area (that is the most beautiful euphemism of the day), we decided to move forward with stakeholder talks to improve the situation. Judiciary Committee held a hearing; similar concerns to what we heard were expressed, but House members were very interested in seeing what Commerce approach could accomplish on a voluntary basis before considering legislation.
Initial meeting of multistakeholder forum on notice and takedown has occurred—archived if you want to listen. Clear messages to group: the topic before us was not legislative change, but improvement within context of current law. True multistakeholder driven process, not decided by USPTO. Inclusiveness and transparency are key goals. We want to see an outcome by the end of the year—best practices, memorandum of understanding. Success is (1) establishing a constructive process that enables real discussion among stakeholders and (2) making some improvement. Initial meeting was setting agenda: consensus on a way forward. Periodic public meetings every 6 weeks, alternating between DC and California. Looking first at standardization of notices, delivery, and processing. Next meeting, May 8. Focus on challenge of small/medium size enterprises; will establish a smaller working group involving self-selected representatives of different constituencies, looking for people with practical/technical operational expertise. Chatham House rules: no identification of who said what, but outcome publicly reported. Larger group will be empowered to take decisions. Nothing will be decided until everything is decided.
Mitch Singer, UltraViolet (I’m sure Singer’s an excellent lawyer, but I can’t see the name of his institution without also seeing red at my horrible Veronica Mars experience, in which the promised digital download on opening day turned out to be, after I finally, finally reached the overloaded website, a digital stream locked to the UV platform, which then didn’t work despite multiple tries. I ended up with a refund and an increased contempt for anticircumvention measures that interfere more with legitimate, paid consumption than with people who want stuff for free.)
People who believe copyright bars innovation see it differently than the rightsholder who says copyright is a property right; all these views are legitimate. §512 struck a balance. Disruption: new tech, new court cases.
Paramount has sent out 75 million takedown notices in the last month; add up all the notices and ask whether it’s really impacted illegal hosting sites’ ability to deliver infringing content. Notices are inefficient and insufficient. Even in whack a mole sometimes there’s no mole, but in cyberlockers and P2P, content is often uploaded faster than we can send notices. The question is: is this really working? While it may be working here, it isn’t working in int’l territories that lack enforcement mechanisms. EU doesn’t required actual knowledge. Instead, reasonableness: ISP/cloud service is in the best position to protect against infringement, and reasonable measures are required. Much more flexible (for whom?) and tech neutral (for whom?) and makes a lot more sense given today’s filtering tech. Content ID: YT puts content in its database and stops uploads. That’s lots better than notice and takedown. It places some burden on a locker, but is much more efficient/less costly.
ECJ confirmed that blocking infringing sites is a balanced approach to protect consumers’ fundamental rights. Probably you don’t think that’s right for the US (SOPA/PIPA), even if the ISP isn’t perfect, if it seriously discourages access to illegal websites it’s the right balance. Rightsholders like that, recognizes that ISP is in best position to stop the infringement (how do you know what’s infringing?); doesn’t allow ISP to turn a blind eye.
Until 2012, we had difficulty getting red flag acknowledgement—courts shut the door with Perfect 10, YouTube, and Veoh. Post-2012, he’s encouraged that the pendulum is swinging: MP3Tunes jury found that Michael Robertson was liable for copyright infringement, because he turned a blind eye to piracy—red flag knowledge. That changed the way we think about level of proof we need. Two other cases: Hotfile and Isohunt—focused on red flag knowledge. Maybe things are shifting. Even in Grokster Court had to reach to patent inducement to find liability against a bad actor.
What should we do in the digital age? (Insert repeat snark about DMCA.) If we knew then what we knew now about filtering, would we have required filtering? Are we going to look at the EU approach and put a heavier burden on ISPs? Are we going to recognize that they’re in the best position to stop infringement? Are we going to require filtering like they do on iTunes and other UGC sites? Are we are going to look at blocking infringing sites? What are we going to do about foreign infringement? What are we going to do about digital first sale? Digital rights portability?
If he had to guess about opening up safe harbors, it will sway in favor of rightsholders because of tech advantages today not present in 1998 (RT: if you have enough money/design capacity/don’t worry about infringing someone else’s filtering patent/etc.).
Michael Carroll, American University, Washington College of Law
Legal aesthetician in him would address: (1) identify theory of liability from which you’re given safe harbor protection; (2) verbs don’t align with exclusive rights—9th Circuit needed 5 pages to explain what rights “storage at the direction of the user” came up with. §512 hallmarks of last-minute legislative deal, but courts are accustomed to providing interpretation to that. What’s the proper allocation of jobs between Congress, courts, and Copyright Office?
Any revisiting could be worse than what we have; private dealmaking going on now may make revisions unneeded. (They’re never going to stop complaining about foreign sites, though.)
Theory of the framework, which was wrong: The deal was: we need to make the internet safe for content. Rightsowners will sit on their content, won’t make it available digitally, unless/until internet is rendered safe. We’ll lock down the internet/have notice and takedown for leaks. But that’s the wrong measure of success. Do we see the progress of science & useful arts? Do we see people creating and people distributing? That should be the measure of whether safe harbors are working. And from that perspective the deal is working out pretty well.
Who knew in the 1990s? It depended on where you looked. §512(c) did in fact have the digital landlord in mind. But the idea that UGC business models weren’t contemplated comes from people who weren’t paying attention. GeoCities was on the radar, and GeoCities was YouTube: we give you space, we give you a template, and we run ads on your website. That’s YouTube. Litman’s Digital Copyright has more on the deal.
We just heard about takedowns being ineffective. Ex parte relief plus DRM was supposed to make content safe; Paramount can’t keep its movies offline. We can stipulate that’s true. But what we need to know is whether a legislative fix is needed. 75 million takedown notices: is that inefficient? Prima facie too big an amount? We now have robots sending notices and robots automatically taking content down. And we’ve heard the story of disproportionate effects on small authors.
Response: we’re in early days. (I’d also say that 75 million ought to be compared to the number of links Google indexes.) Also we hear that takedowns are unfair because Google is making too much money. But: Would you have gotten Content ID if Congress had drafted the requirement for what Content ID should look like? He submits that you would not.
Others say: Maybe we should have a sunrise period for owners to inspect content before it goes online. Or maybe we should require filters. These are the sort of proposals. Or some sort of burden shifting on staydown/generalized notice. This shifts balance in favor of rightsholders and not public interest. We are seeing an explosion of creativity, even if certain highly capitalized business models are unable to adapt; progress is not under threat and there’s no reason to reopen this particular deal especially given the kind of institution that Congress is.
Real problem: failure to license. If more Paramount movies were on Netflix, they’d be sending fewer takedown notices. (I note he said nothing about more Paramount movies on UltraViolet.) Netflix is not “unsafe,” but they can’t agree on the price of content on demand. Consumers are Roger Daltrey: “I want it.” Rightsholders are Pete Townsend: “You can’t have it.” Consumers should and ultimately will prevail.
Are we getting enough creativity? Look at Frozen. We are getting smash hits; we are getting big investments.
What of small producers who find their works on YouTube? That’s a problem. Next YouTube startup is also a small producer, and shift in §512 penalizes them. Small/independent authors are put up in the legislative conversation but they’re not the real drivers of reform. They’re better off with the battle of the titans, which will lead to a licensing scheme they can use too.
If we did do something, we should address abusive takedown notices, giving users right to use lawful use/fair use flag constituting preemptive counternotice, requiring direct suit against the uploader and preventing suit against the ISP. Not everyone will use, but will have powerful educational effect.
Brianna Schofield, UC Berkeley School of Law, Samuelson Law, Technology & Public Policy Clinic
Need for further research: difficult to know how §512 should change, if at all, if we don’t know how it’s actually working. Takedownproject.org: go to our website. A lot of hidden decisions on both sides: senders and recipients. Recent and growing automation has led to vast amounts of data. We’ve begun collecting and coding. Coding whether the notice requirements are met; who’s the owner; how many links per notice; how the allegedly infringing work is identified. Designed to tease out answers about validity, whether it’s targeting something that’s not a copyright claim but might be something else, like a privacy claim. Transparency has been an uphill battle; ISPs have been hesitant despite confidentiality promises because of fear of scorched earth litigation. Rights enforcement agencies sometimes use transparency as a portfolio of their work.
We have been asking ISPs about form notices. Some say this leads to a significant decrease in notices, while others say dramatic increase. Overall, form notices do seem to lead to greater compliance with statutory requirements, but not clear if that relates to underlying validity of claim.
Notice and staydown: Not practically/financially feasible for many ISPs. Content ID took hundreds of engineers and $60 million to build; not all ISPs can afford that. Competitive disadvantage; effectively lock out new and innovative services. Filtering doesn’t account for things like fair use, licensed use, etc.
What about abusive notices? Notices are used to target competitors’ content, critical speech, claims masquerading as copyright but really about TM or privacy. The counternotice provisions are ineffective. Even problematic notices: the safest thing for an ISP to do is comply. Once they do that, they notify the users, but that’s super intimidating, especially for those who wish to engage in anonymous speech. Insufficient legal disincentives for senders. §512(f) is expensive, timeconsuming, with little prospect of damages.
ISPs still maintain that §512 is vital: woven into how the internet functions. Reform must not disrupt benefits of system.
Bridges: from the trenches, the safe harbor is not safe once you get into litigation. It’s great when the system works. Perlmutter says designed as alternative to litigation; but if one has to litigate the safe harbor, life is hell. Even though you should advise an ISP to do it as the first thing, it should be the last thing litigated, because defending against every theory of copyright infringement is too hard. D can win sj by winning three issues: direct (no volition); contributory (no intent); right and ability to control (vicarious). If you win those you win the case without safe harbor. If you litigate the DMCA you have to win on eight issues, which is a ton more expensive. Many of the biggest cases have taken an extraordinary financial toll on defendants—Google is rumored to have spent over $100 million defending itself. Veoh: the company that did it all right, complying with DMCA and signing voluntary MOU on UGC principles. The idea was a peace treaty. But Veoh got sued, and it won a significant flat-out victory on the DMCA, and went bankrupt winning the case despite starting with a lot of money. If you talk about a voluntary system in which “stakeholders” agree, how will that protect anyone from expensive litigation by aggressive copyright owners?
Perlmutter: she doesn’t know much about Veoh and won’t express an opinion on the outcome, but voluntary agreements aren’t intended to be a shield against litigation, though if they can be that’s great. (So what are they for? Staving off legislation?) Only legislation is a shield, and even then there’s no guarantee, because there won’t be clear bright lines because it will always be a balance. Best practices approach: meant to be a model. Courts might look at that, but no one will be bound. Or binding agreement between particular entities, such as the Copyright Alert System, which could include promises not to sue each other if there’s compliance. Green Paper: stakeholders will decide what type of output there will be and whether it will be best practices or binding, but she’d guess it will end up closer to best practices given diversity.
Clarification: we are not in the Green Paper process endorsing a call for standardization, but asking whether it could be beneficial.
Bridges for Singer: you were concerned about actual knowledge as too lenient for ISPs. Safe harbor comes in only in damages analysis—you’re still entitled to relief. If we take actual knowledge away, what is the remaining underlying standard for contributory infringement, where Sonysaid constructive knowledge wasn’t enough? Should Sony be overruled? (Grokster may have changed the standard from knowledge to intent, but assuming knowledge is still in play, then what?)
Singer: He doesn’t know. He’s not making a particular proposal. There’s something broken with notice and takedown, and there’s a better way (that he doesn’t know). Not sympathetic when someone knows that the majority of content posted will be illegal and then says they have no obligation other than to respond to notices. Can’t go into detail about requisite knowledge, but at some point when you receive notice after notice for the same movie that’s in theatrical release, there has to be something more. Tech exists to take reasonable measures. Other panelists think notice is greatest thing in the world (not quite) but there has to be a better way.
Carroll is odd to say that because people are stealing your content and you can’t stop them, you need to change your business model. No one should change a business model because people are stealing their stuff. That’s not the right kind of disruption. (Well, that depends on the alternatives, is I think the point.) We wouldn’t get to show films in theaters if we had immediate online distribution. The response should be “how can we work together to stop trafficking in stolen goods.”
Carroll: his point is descriptive. You can’t stop the internet, whether you like it or not. If you don’t shift the business model, you will get a higher level of unauthorized use than you would in a different model. I’m not telling you to change, I’m telling you about unsatisfied demand.
Bridges: SOPA was called “stop online piracy.” But there have been at least a dozen laws since 1982 whose purpose was to “stop” piracy. What was the metric for “stopping” piracy? What is the acceptable level without collateral damage to the broader ecosystem of speech? Microsoft’s goal: it should take more than 20 minutes to find an illegal copy.
Singer: we won’t stop online piracy. His metric is his son, who says he can watch a movie within seconds. That’s not the kind of metric he likes to hear. Difficulty! Accepting notice and takedown isn’t enough. Why not have a filtering standard, and if you filter you get better treatment? The environment is different for us than for music, because we don’t release movies except in theaters for the first four months. The source of that content online is taken from a movie theater, not from someone’s personal copy. We can make progress just by focusing on that content.