Background

Under current law, ISPs can be held liable for their customers’ infringement under theories of secondary liability. Before 1998, the basis for this liability was grounded in copyright law’s contributory and vicarious liability doctrines. However, ISPs complained that the courts did not consistently apply these doctrines. As a result, while some courts exempted ISPs from liability for their customers’ acts of infringement, others held ISPs liable even when they had no knowledge of the infringements. At the same time, the music and movie industries wanted ISPs to bear at least some responsibility for their customers’ infringements. The Digital Millennium Copyright Act (DMCA) was passed in 1998 to address these concerns. It provided safe harbors from secondary liability to ISPs and online services that met certain conditions. These safe harbor provisions of the DMCA are contained in section 512 of the copyright statute.

Under Section 512, ISPs – companies that provide access to the Internet - are exempt from infringement liability, provided that the ISP in question maintains a policy of terminating access for repeat infringers. The idea is that an ISP, like AT&T, Verizon, or Earthlink, is a passive conduit that cannot control the information that flows on its network. Online Service Providers (OSPs) – which range from hosting services such as YouTube to search engines such as Google Search, message boards and others - are also exempted from liability, as long as they take down infringing material upon notice from the copyright owner. If they do not meet these takedown obligations the OSPs can face lawsuits for copyright infringement.

Chilling Effects

Section 512 has succeeded in securing OSP co-operation in taking down infringing materials from hosting sites and search engines. Since OSPs face no liability for taking down content, the law has encouraged takedowns even when it’s not clear whether or not a work is actually infringing. But in doing so, section 512 has also had a chilling effect on free expression online. One study found that the notice and takedown procedures have been used in ways not intended by the statute. It has been used in some cases to thwart competition and in others by people who don’t even own a copyright. For instance, some businesses have sent notices requesting that search engines remove links to their competitors’ websites. In others people who appear in photographs have sought removals even where they did not own copyright in it.

Surprisingly, the aforementioned study found that the movie and music industries that pushed for these provisions have rarely sent notices to companies like YouTube and Google. Rather, they have preferred to send notices directly to ISPs. The purpose of these notices might be to build a record of infringement against individual customers so that the ISP will be obliged to terminate their Internet access. The study notes that the primary targets of these notices are P2P filesharers.

Finally, while takedown notices are sent out by content owners on a daily basis, one little known fact is that there is actually a “put back” option for users. This option is available if content is taken down by mistake or based on a misidentification of the material in question. However, the law makes it procedurally more difficult to “put back” than to takedown. Therefore, putback rights are rarely exercised.

Filtering

Not satisfied with the section 512 provisions, representatives of the major music and movie industries are now pressuring ISPs and OSPs to actively monitor their sites for infringement and to use filtering technologies to prevent infringement from occurring in the first place.

Viacom's Lawsuit Against Youtube

Viacom’s lawsuit against YouTube is an example of this kind of pressure. In its $1 billion lawsuit, Viacom claims that YouTube serves as a means for uploading and disseminating a massive amount of copyrighted content. Viacom claims that YouTube and its parent company, Google benefit financially from this copyright infringement and while the company should be taking active steps to prevent it, YouTube simply looks the other way, in its own financial interest. Google, on the other hand, claims that it complies with the section 512 requirements and has already removed 100,000 clips at Viacom’s behest. In the suit, Viacom asks for the removal of an additional 150,000 clips. The larger problem with this lawsuit is that it questions the legality of all websites that allow users to upload their own content.

Is All Content Uploaded to YouTube Infringing?

While many clips uploaded to YouTube are infringing, a study by the American University Center for Social Media reveals that a large amount of purely user-generated content is also uploaded to the site. Additionally, many of the clips that feature copyrighted content are actually making fair use of that content. For example, MoveOn.org’s parody of the Colbert report, "Stop the Falsiness" was one of the 100,000 clips taken down at Viacom’s request. In response, the EFF filed a lawsuit against Viacom, seeking a ruling from the court that the clip was not infringing.

YouTube's Announcement to Filter

Meanwhile YouTube has introduced a beta version of a proprietary video identification system that would prevent infringing content from being uploaded to the site in the first place. The system works by matching user submitted videos to a reference database of copyrighted works. If the user's video matches content in the database, YouTube will block it or take any other action that the copyright owner requests. Such action may include requiring the user to upload the video only if he pays a license fee to the copyright owner. The user is permitted to object to a block and if she does so, the notice and takedown provisions of section 512 take over.

Public interest groups are concerned that this system will adversely affect the posting of fair use videos. To address this concern, they have suggested a 3-strikes method that would make automated copyright filtering more consumer-friendly. Under this method, content should not be removed unless it has “three strikes” against it. Here's an example involving a video clip that is believed to be infringing. Strike one: the video track in question matches the video track of a copyrighted work. Strike two: the audio track of the clip matches the audio track of the same copyrighted work. Strike three: nearly the entirety of the track (90% or more) consists of the same copyrighted material. If all these criteria are not satisfied, the clip should be viewed by a human being, in order to determine whether or not the clip infringes on a copyright.

Network Filtering

While the Viacom lawsuit focuses on a content host, there is a larger movement to require ISPs to filter their networks for copyright infringement. On June 15, 2007, NBC Universal filed a petition asking the FCC to require broadband providers to “use readily available means to prevent the use of their broadband networks to transfer pirated content.” In November 2007 AT&T announced its intention to filter its networks, again to prevent copyright infringement. Meanwhile, bills have been introduced in Congress pressuring universities to filter their networks for copyright infringements.

How it works

Currently, ISPs may use two types of filtering technologies: content inspection and traffic analysis. Content inspection involves looking at actual packets of information traveling over the network (a practice that is often referred to as "Deep Packet Inspection" or “DPI”, for short), regardless of whether the information in the packet pertains to an email, a P2P file or any other application that sends data over the network. The ISP will then have to compare the content in question to an existing database of copyrighted content and must decide whether to block the content or let it go. Of course, this comparison is often done electronically, without any human intervention. Traffic analysis, on the other hand, involves analyzing the type of traffic (i.e. whether the data in question pertains to an email or a P2P file) and then blocking any traffic that is believed to infringe copyrights.

Problems with Filtering

Regardless of the technology used, copyright filtering presents several problems. First, filtering is both an inefficient and potentially harmful tool when it comes to protecting copyrights, as filtering technologies tend to be simultaneously overinclusive and underinclusive. Filtering technologies are overinculsive because they often filter legitimate content along with infringing content. For example, if an ISP decides to delay, block, or degrade all traffic relating to a P2P application (as Comcast did with BitTorrent), that ISP will block all of the legitimate uses that application has. P2P applications like BitTorrent are often used by independent musicians and open-source software developers as a free distribution platform for sharing their work online. P2P protocols are also used by, video services like Joost and Miro to distribute video content legally. Even if the filters look closely enough to determine whether or not the content is infringing, content filters are often unable to distinguish between an actual infringing piece of content and a piece of content that is protected under the doctrine of fair use, for example, a backup copy of a work, a space/time-shifting copy, a mashup or excerpt or a me-to-me transfer.

Meanwhile, filtering technologies also tend to be underinclusive. No piece of technology will ever work perfectly and copyright filters will surely fail to identify some copyrighted content. What's more, once such filters are widely implemented, enthusiasts and hackers will quickly develop methods for circumventing these filters. With current methods of filtering, a technique as simple as sending data over an encrypted channel can be enough to avoid the filter.

So, what’s the solution? While copyright owners have a right to protect their content, it is equally important to protect the consumers fair use rights. Content inspection poses a serious threat to privacy by examining contents of communications like emails. Traffic analysis could repress innovation in distribution technologies (for example, P2P protocols) which evolve over time by blocking certain protocols. For this reason, we feel that consumer education, alternative licensing schemes and improved offerings by content providers online are all better solutions to the problem of compensating artists for files traded online.

Conclusion

Copyright owners’ expectation that ISPs help them protect their rights is not unjustified. But any protection mechanism must be based on a balance between the copyright owner’s right to protect his work and the user’s rights. Users’ rights include procedural fairness such as being notified before content is taken down, and being given an opportunity to protest that take down. Some have suggested that the notice and take-down procedures be amended so that user’s are notified before content is taken down and offered an opportunity to provide a counter notice. Also, Public Knowledge has suggested as part of its six point program for copyright reform that copyright holders who “knowingly or recklessly” send false take down notices should be punished. Current law only punishes copyright holders for “knowingly” sending false notices.

In addition, rules to prevent infringement should take to into account that the Internet is an essential means of communication in today’s world. It is used for myriad purposes including banking, shopping, telephony, exchange of scientific information, expression of political views and much more. Industry moves to prevent copyright infringement on the Internet through filtering techniques should be monitored by the government to prevent abuse of the basic right to Internet access.