Establishing an Indirect Liability System for Digital Copyright Infringement in China: Experience from the United States’ Approach

By Xiao Ma* Download a PDF version of this interview here.  

Introduction

Indirect copyright liability is a term used to describe the liability imposed upon a defendant who is not the direct infringer,[1] but whom the law nonetheless holds liable for damages which the copyright owner suffers from the infringement.[2] Indirect liability requires that the defendant (also referred to as the “indirect infringer” below) assist, promote, facilitate or benefit from the direct infringement.[3] The indirect copyright liability rule has different names such as secondary liability or derived liability,[4] and has various forms across jurisdictions.[5] For instance, the United Kingdom adopts the notion of “authorisation” in its copyright legislation to deal with intermediaries who authorise the infringing acts done by the third party.[6] The United States, on the other hand, has developed doctrines of contributory liability, vicarious liability and inducement liability through case law.[7] In China, however, there has been no notion of “indirect liability” in the copyright law regime. The issue of indirect copyright liability has been dealt with under the joint tort liability regime.[8] In order to address digital challenges, especially the increasing use of peer-to-peer (P2P) technologies, China has developed a series of instruments—including laws, regulations, judiciary interpretations and administrative measures—to regulate Internet intermediaries’ copyright liability for infringement committed by users. These legal documents establish a joint tort liability system under which network service providers (“NSPs”)[9] share joint liability with direct infringing users under certain conditions. The most notable of these legal instruments includes: the Copyright Law of the People’s Republic of China 1990[10] which was revised in 2001[11]and 2010;[12] the 2000 Networks Copyright Interpretation issued by the Chinese Supreme People’s Court, which was amended twice in 2004 and 2006;[13] the Regulations for the Protection of the Right of Communication through Information Network, which embodies a set of so called “safe harbor” provisions;[14] the Tort Liability Law of the People’s Republic of China[15] with one specific provision regulating ISP liability;[16] and the Provisions of the Supreme People’s Court on Several Issues Concerning the Application of Law in Hearing Civil Dispute Cases Involving Infringement of the Right of Dissemination on Information Networks.[17] However, in China, debates surrounding intermediary liability issues have never stopped. First, the judiciary and scholars have been confronted with an obscure legal underpinning of indirect copyright liability due to the lack of an independent category of indirect infringement in the tort law. The success of an indirect copyright liability claim under the joint tort liability principle without the existence of joint intent or knowledge is questionable. It is equally problematic to impose indirect liability on a separate tort basis. To overcome this lacuna, a survey of current tort law is required to test the viability of engaging indirect copyright liability under the guidance of the general tort law regime. Second, courts have been struggling with the incompleteness of liability attribution rules, especially the constitutive element that has been used to establish the culpability of the defendant. The issue in the spotlight is the inconsistent knowledge standard. Current rules that are used to define and interpret whether someone “knows,” “has reason to know” and “should know” have been the subject of contention. Third, the Chinese safe harbor provisions, inspired by the United States DMCA safe harbors, fail to accommodate the specific conditions in China. The 2006 Regulation adopted the legislative model of ruling on both liability attribution and liability exemption, which caused confusion as to the nature of the safe harbors. Unlike the United States’ approach that has provided specific guidance on the interpretation of the safe harbors through case law, Chinese legislation has failed to correctly address the knowledge standard or control and benefit standard. In addition to addressing the above-mentioned debates, this article thoroughly reviews current laws and cases in China and compares them with United States laws on indirect copyright liability. In the United States, liability attribution rules have evolved and formed three categories: contributory liability, which stresses elements of constructive knowledge and material contribution; vicarious liability, which emphasizes factors of ability to control and direct financial interest; and inducement liability, which focuses on purposeful, culpable conduct and expression. China has adopted the aiding and abetting liabilities which correspond to the United States’ contributory and inducement liabilities. However, the one-sided emphasis on applicable knowledge has impaired the importance of other constitutive elements such as the intent, the relationship with direct infringers, the ability to control direct activities, the means that are used for infringement, and due care. The United States has developed concrete explanations for the application of exemption rules in indirect copyright liability in case laws.[18] In China, however, safe harbors have set a higher standard to exempt liability for NSPs, compared with the standards of the United States.[19] This article begins by exploring three major debates concerning Chinese NSP liability rules. The issues in these debates are factors that have hampered the establishment of an indirect copyright liability system in China. Through an analysis of recent efforts that China has taken to construct a set of NSP liability rules, this study highlights the incompatibility of safe harbor rules, provides an overview of comprehensive liability factors, and calls for remolding and updating China’s relevant regulatory instruments by reexamining the United States’ experience. Part II revisits the key doctrines of indirect copyright liability in the United States, finding that judicial practice has stepped out of the traditional analysis of knowledge and material contribution elements in contributory liability and of the control and financial benefit factors in vicarious liability. Instead, a series of subjective and objective factors—including the relationship, control, knowledge, means and due care—have all invited consideration. Part III proposes guidelines and specific suggestions in establishing an indirect liability system for digital copyright infringement in China. This article concludes with a strong recommendation that, to construct an efficient, well-balanced and predictable indirect copyright liability system for dealing with digital copyright infringement in China, an independent tort theory-oriented, culpable conduct-based indirect copyright liability system, with modified safe harbor provisions, is needed.

I. Debates on NSP Liability Rules in China

In China, there have been no systematic rules on the issue of indirect liability for copyright infringement in legislation. This was compatible with the original focus of copyright protection in China, because the legislative and applicative focus had been the act of direct infringement since establishing the copyright system. This focus can partly be explained by the previous technological conditions where the general public could not afford to copy technologies and only a select few, with certain financial and material resources, were capable of conducting severe direct infringement. In this light, the copyright owners could gain sufficient remedies through chasing legal liabilities from those direct infringers. With fast developments in copying and disseminating technology, however, individuals are now capable of conducting severe copyright infringement. Nevertheless, it is not easy for copyright owners to locate the infringers, nor to gain sufficient compensation from chasing individual liability. Moreover, if they are not immediately restrained, some expanding infringements will produce very serious outcomes for copyright owners. In practice, the uniformity of court judgments at different levels is not guaranteed.[20] Therefore, the copyright laws in China should not only introduce fundamental principles guiding indirect infringement of copyright, but also make concrete and detailed provisions to construct a comprehensive and systematic indirect liability system for copyright infringement online. Civil liability for copyright infringement is stipulated in relevant articles of the General Principles of the Civil Law (“GPCC”), which prescribes that “citizens and legal persons shall enjoy rights of authorship (copyrights) and shall be entitled to sign their names as authors, issue and publish their works and obtain remuneration in accordance with the law.”[21] When dealing with infringement, it provides joint liability for joint infringements.[22] The Supreme Court’s Opinion of the Civil Law further states that “any person who incites or assists another to commit a tort is the joint tortfeasor.”[23] Relevant articles in the 2010 Tort Liability Law can be applied when dealing with copyright liabilities,[24] including strict liability, fault-based liability, and joint and several liabilities.[25] It is noteworthy that the 2010 Tort Liability Law prescribes defines abetting and assisting activities as two forms of infringing acts which lead to joint and several liabilities.[26] Article 36, which specifically deals with NSP liability in the network environment,[27] affirms the notice and takedown provisions[28] provided in the 2006 Regulation[29] and makes clear that a fault-based liability is applied to the NSP who has knowledge of a third party’s infringement.[30] This article is a principled stipulation for indirect liability that needs to be further interpreted. The main problems involve the interpretation of certain terms in this article such as “necessary measures,” “in a timely manner” and “know.” For example, whether “know” is perceived as actual knowledge, or as knowledge including “should know,” was under hot debate.[31] The recently issued 2013 Provision has absorbed recent judiciary practice as well as recent academic research.[32] By distinguishing direct and indirect infringement of network service providers,[33] especially stipulating the abetting and aiding infringement[34] with a specific knowledge standard,[35] the 2013 Provision has provided a consistent framework for evaluating and assessing NSP infringement with regard to the relevant issues over online copyright disputes. The imputing and limiting of online copyright liabilities are embodied and the balancing of interests is reflected. Based on the investigation of cases and consultation from academes and different sectors, the promulgation of the 2013 Provision will help courts at all levels in China to keep pace with the times and be more scientific and normalized in adjudicating copyright cases in the future. However, for several reasons, the 2013 Provision does not provide total clarity on attribution rules. First, it has been established that fault is a determinative factor in deciding whether the NSP is liable for infringement as an abettor or aider.[36] However, the 2013 Provision defines fault as knowledge, which overlooks other types of intentional fault including inducing or encouraging.[37] Second, the 2013 Provision takes an approach that stresses only the subjective element rather than the objective element. Knowledge (know or should have known)[38] becomes the key deciding factor; the objective factors, such as nature of the services provided and the NSP’s capability of information management, are treated as factors to be taken into account in deciding the knowledge.[39] Third, the factors in deciding fault of an NSP are enumerated with examples, which limit judges’ discretion.[40] This legislative technique becomes disadvantageous when facing new technologies. Last but not least, the provisions relating to copyright liabilities in various legal instruments have not yet been unified, especially considering that the Copyright Law is currently under the third revision. Therefore, the effect of application of the 2013 Provision remains to be seen.

A. Debate on the Tort Basis for NSP Copyright Liability

Generally speaking, tort infringement acts lead to tort liability. However, under current tort theory, debates among scholars on the tort infringement basis of NSP copyright liability have been getting heated. Some argue for joint infringement,[41] while others argue that NSPs are severally liable.[42] Prior to the promulgation of the 2010 Tort Liability Law, the act of joint infringement was used to define the relationship between the NSP and the network users; for instance, the 2006 Network Interpretation provides that the NSP shall bear joint liability with the users.[43] The same approach was adopted in Article 23 of the 2006 Regulation.[44] However, the 2010 Tort Liability Law recognizes joint and several liabilities for aiding and abetting torts, without referring to its liability basis.[45] In judicial practice, Chinese courts have been employing the joint infringement theory. For instance, in the case Zhongkai Culture v. Guangzhou Shulian[46], the court applied the 2006 Network Interpretation, stating that the defendant, who did not upload a film for direct downloading, had abetted and assisted the users in committing the infringement and should bear joint infringement liability with direct infringers.[47] Among academics, Professor Ming Yang argues for joint infringement as a basis for NSP liability in China.[48] He points out that unlike the criminal law, joint infringement in civil law tort theory stresses an objective infringement act, rather than the existence of conspiracy or joint intent or knowledge.[49] If the indirect infringement system is used to define NSP liability in the form of assisting infringement, major confusion will occur surrounding the nature of the relationship between a direct and indirect infringer.[50] Based on this observation, he claims that it is unnecessary to import an indirect infringement system in China.[51] However, the above point of view has been challenged by Wei Xu, whose argument stems from the concept of joint infringement itself.[52] The major disagreement concerns the question of whether the infringements committed by multiple persons without connected intent constitute joint infringement. For a start, he suggests a systematic reading of articles 8 through 12 of the 2010 Tort Liability Law, and finds that the answer to the above question is no.[53] He advises assisting infringement liability rather than joint infringement liability in a situation without connected intent between NSP and its users.[54] Next, he analyzes a dilemma in practice if the NSP is considered as a joint tortfeasor. Under the joint infringement theory, the joint tortfeasor is required to participate in the necessary joint action as a necessary party to the litigation.[55] However, in practice, users who commit direct infringement are highly impossible to locate, or the cost is too high to be added as a party to the litigation.[56] In addition, the NSP that bears liability has a right to pursue recovery from direct infringers, but it is difficult and impractical for an NSP to do so.[57] Accordingly, Xu proposes that the NSP shall be severally rather than jointly liable.[58] Tort law Professor Lixin Yang provides a third opinion.[59] Professor Yang describes a logic gap between the infringement committed by multiple parties and respective liability forms in traditional tort law theory in China.[60] The unresolved question is whether the combination of direct infringement and indirect infringement constitutes joint infringement. He advises that it is not joint infringement but joint and several liabilities based on consideration of public policy.[61] From his point of view, infringement activity leads to corresponding infringement liability. Given that, joint infringement leads to joint and several liability, the independent infringement act leads to proportionate liability, and the third party infringement leads to third party liability. A gap, consequently, is generated concerning the correspondent form of infringement, which leads to unreal joint liability.[62] He proposes a concept of “overlapping infringement,”[63] constituted by multiple infringement activity, which has direct or indirect causation of the damage. The infringer bears the unreal joint liability. When the concept of overlapping infringement comes into play, the logic gap that has long existed in tort law theory is filled. In this light, the theoretical basis for NSP liability can be initially established.[64]

B. Debate on the NSP’s Knowledge Standard

The requisite level of knowledge possessed by the NSP in determining its liability of online copyright infringement committed by its users has also been one of the major controversies in China. The 2000 Network Interpretation raises the “explicit knowledge” requirement in Article 5.[65] The 2006 Regulation sets forth the requisite knowledge requirement for limiting the liability of NSPs that provide services related to information storage and searching or linking. Article 22(3) employs the terms “knows” or “has reasonable ground to know” for limiting liability of an NSP that provides storage space,[66] while Article 23 uses the phrase “explicitly knows” and “should have known” for an NSP that provides searching or linking services.[67] Prior to the promulgation of the 2013 Provision, it was unclear what “should have known” means. For instance, compare IFPI v. Baidu[68] and IFPI v. Yahoo.[69] Both cases are very similar in that they concern recording labels suing search engines and linking service providers. Nevertheless, they produced different results. In IFPI v Baidu, the IFPI sued Baidu on behalf of seven music companies for providing links to a website that offered free downloading of infringing songs.[70] The Beijing First Intermediate People’s Court ruled in favour of Baidu, holding that the plaintiff failed to notify Baidu of the infringing files, and Baidu had no fault for searching and linking to the other websites because it ‘should not have known’ of the infringing material.[71] The Beijing Higher People’s Court upheld the decision. However, in a similar case IFPI v Yahoo, the defendant Yahoo who provided links to the third websites for free music downloading, was held liable for aiding infringement.[72] It is notable that though the defendants in both cases provided the same service, there were different facts that were essential for the decision. The music labels sent notices to the defendant including the name of the song, the name of the album and the singer of each infringing file. The court held that the above three pieces of information were enough to locate the copyrighted music However, Yahoo refused to disconnect the access to the infringing material.[73] Yahoo should have known of the infringement, therefore it constituted aiding infringement.[74] In another case Fanya v Baidu,[75] the right holder also sent notices to the defendant, demanding the defendant to disconnect access to the infringing material. However, these notices were different from the notices sent in IFPI v Yahoo that were sufficient for locating the illegal music files. Instead, the notices from Fanya only provided the name of the song, with which alone would be very difficult to locate the copyrighted material enjoyed by the copyright holders.[76] In order to understand what constitutes “know” or “should have known” for a searching or linking service provider, first, a notice that is sufficient to locate the infringing material is required. An noncompliance notice will not constitute explicit knowledge; but, if it is sufficient to locate the infringing material, the “red flag” test is met.[77] Second, there are different types of service provided by the searching or linking NSPs. One is the search box, which has no bias towards the content it linked when the key words are being typed in. The other is the list provided by some search engines. By clicking song or singer’s names in the list such as ‘top 100’ songs, users can reach the same result as typing the name in the search box. However, the culpability of the search engine provider under the two situations is different. By providing the list, the search engine providers are supposed to have higher duty of care. When infringing material is apparent like a ‘red flag’ from a glance of the list, the service provider is required to disconnect it, rather than turning a blind eye to the infringement.[78] The NSP knowledge standard has also been uncertain since the drafting of Article 36[79] of the 2010 Tort Liability Law.[80] There is no doubt that “know” includes “explicitly know.” However, whether it also includes “should have known” or “have reason to know” generates fierce debate. Some legal professionals maintain that “know” includes “should have known” because “many culpable internet service providers might escape liability if their liabilities are only based on actual knowledge,”[81] while some disagree because “it would incur a considerable duty of care for website operators.”[82] The third opinion suggests that the knowledge requirement includes “have reason to know” instead of “should have known,” because the latter may increase the level of care for internet service providers, whereas the former is equal to “awareness of facts or circumstances” under the DMCA.[83] Though the two terms both relate to the term “constructive knowledge,” and vaguely to the duty of care under the tort of negligence at common law, they have a slight difference:

The term “had reason to know” has a slight difference with “should have known”. The first centers on the “reason”, while, the second emphasizes the “duty” which could be a legal duty of care under a statute or based on an agreement. The “has reason to know” is more like a factual determination based upon the circumstances and information available to the parties, while, the “should have known” is more like a judicial determination of what is a reasonable level of knowledge given the parties and the circumstances. Hence, the first is a party-specific objectivity, and the second is a community-focused objectivity.[84]

Tao disagrees with all above opinions in interpreting “know” in the 2010 Tort Liability Law and instead proposes the United States’ approach in the DMCA.[85] Since China has enacted no rule preventing the imposition of a monitoring obligation,[86] the first and third approaches would result in disproportionate burdens on intermediaries. In addition, Tao argues that the adaptation of the term “reason to know” and “should have known” in Article 22 of the 2006 Regulation is a misunderstanding by legislators[87] and that an interpretation consistent with the DMCA would avoid more confusion and legal uncertainty.[88] Prior to the promulgation of the 2013 Provision, Tao’s interpretation of “know” in Article 36 of the 2010 Tort Liability Law was of great significance, because a broad interpretation of the knowledge standard would lead to an excessive burden for the NSP. However, since 2013, the status of NSP liability has changed. Indirect copyright liability attribution rules, which employ “know” or “should have known” have been introduced based on the knowledge of an NSP.[89] Under the rudimentary framework of indirect liability for digital copyright infringement, the knowledge requirement in different legislation needs to be re-examined. Article 36 of the 2010 Tort Liability Law adopts a horizontal approach which is applicable to both intellectual property infringement and other civil claims governing content including copyright, and defamatory and obscene content.[90] It is advisable to interpret this article in a broad way, because it “is not only fatal to the imposition of copyright liability on [NSPs] but also important for other kinds of content-related liability for [NSPs] such as liability for defamatory content.”[91] Further, as a liability attribution rule rather than liability exemption rule, the knowledge requirement in Article 36 is obviously broader than that of the safe harbor provisions under the 2006 Regulation. It is therefore concluded that the term “know” in article 36 of the 2010 Tort Liability Law embraces the meaning of “explicitly know” and “should have known” to determine the NSP’s knowledge of copyright infringements committed by the third party.

C. Debate on the Nature of Safe Harbors

There have been passionate and polarized debates on the nature of the Chinese “safe harbor” provisions[92] provided in the 2006 Regulation.[93] A contextual analysis suggests the language used in the 2006 Regulation reflects divergent approaches towards the liability model. For instance, the safe harbors all use the language of liability limitation, such as “shall not be liable for damages”;[94] however, the Regulation provides for liability attribution in the second section: “where [the NSP] knows or has reasonable grounds to know that the linked works, performances, sound recordings or video recordings infringe another person’s right, it shall be jointly liable for the infringement.”[95] Given the ambiguity of what is prescribed in the legislation, both courts and academics suggest clarifying the nature of the provisions.[96] Although the safe harbors have generally been considered by academic circles as liability exemptions, some disagree.[97] There is general international consensus that the purpose of safe harbors is to be a “limitation of liability” or an “exemption from liability.” For instance, the European Union Directive on Electronic Commerce[98] defines the safe harbors as a “limitation of liability” or as constituting an “exemption from liability.”[99] The same approach is adopted in the United States.[100] In China, the 2010 Guiding Opinion issued by the Beijing Higher People’s Court referred to Articles 20 to 23 of the 2006 Provision as constituting a “liability exemption.”[101] Similarly, the mainstream academic view is that the major purpose of the notice and take down procedure is to exempt NSPs from indirect liability for direct infringement committed by network users; that the nature of the NSP safe harbors is not liability attribution but exemption; and that the safe harbors are not the final establishment of liability but the defenses.[102] However, Wei Xu disagrees with the academic consensus.[103] He argues that adopting the theory of liability exemption contradicts the NSP’s fault-based liability for third party copyright infringement.[104] He proposes the following inferences, based on the theory of liability exemption. First, infringement liability is established before the notice is received by the NSP, and the NSP is exempted from liability as soon as it deletes the infringing content.[105] Alternatively, infringement liability is not established before the notice is received by the NSP but after, and as soon as the NSP deletes the infringing content, it is exempted from liability.[106] He indicates that the first inference contradicts fault-based liability because before the notice is received, the NSP bears no fault.[107] The second inference, according to Xu, is also incorrect because if receiving the notice means fault, this would contradict Article 36(3) of the 2010 Tort Liability Law,[108] which suggests that fault means knowledge and failing to delete.[109] However, this inference has severe defects. First, the author falsely states that the fault-based liability requires that before the notice is received there has been no fault with the NSP. This statement equates fault with notice, leaving out the possibility of other forms of knowledge of an NSP including an email from a third party, actual knowledge obtained before receiving the notice, or red flag knowledge.[110] Second, the author misunderstands the meaning of fault as knowledge plus failing-to-delete. It has been made clear under Article 8 of the 2013 Provision that fault of the NSP means knowledge of a network user’s infringement.[111] Qian Wang, a leading professor in the area of NSP copyright liability, argues that the safe harbor provisions are actually two sides of the same coin—serving as both the attribution and limitation of NSP liability.[112] By analyzing each condition in the safe harbor provisions, he suggests that some liability exemption provisions correspond to direct liability and some correspond to aiding liability.[113] For instance, Article 22(3) provides that the NSP shall not be liable for damages if “it does not know or has no reasonable grounds to know that the works, performances, sound recordings or video recordings provided by its subscribers infringe any other persons’ rights.”[114] This is obviously an expression of liability attribution. Expressing this article another way—as “knows or has reasonable ground to know”—forms the attribution of aiding liability. Prior to the promulgation of the 2013 Provision, this “two sides of the same coin” approach facilitated the understanding of the NSP’s copyright liability for infringement committed by users under circumstances for which there were no relevant laws regarding indirect liability. Since the 2013 Provision has been issued, Wang’s approach, which has been adopted by China’s courts for years, is no longer compatible. First, not all safe harbor provisions correspond to a form of liability attribution. Distinguishing these provisions becomes a difficult task when applying the safe harbors. Second, problems occur with applying the conditions of each safe harbor: Are the conditions necessary, sufficient or both? An in-depth analysis of United States’ safe harbor provisions below will facilitate an understanding of the changes in China’s safe harbor provisions. An analysis of laws, cases, and academic debates over recent years in China shows that the absence of systematic indirect copyright liabilities is the root of confusion and chaos in the Chinese copyright regime. The lack of a comprehensive indirect copyright liability scheme not only leaves a degree of uncertainty to the issue of digital copyright infringement, but also creates a loophole in Chinese tort law. Unlike the United States, which has established indirect liability for intellectual property infringement through legislation and the common law, China has been left behind—although a significant body of research has contributed to this area. It is reasonable to speculate that the inconsistency in the copyright regime, even in the tort law regime, has posed a significant threat to the interpretation and application of digital copyright infringement in judicial practice.

II. Experience from the United States

China, as a developing country, has gained some nutrition from the existing body of law in the United States. For instance, China’s 2006 Regulation was a result of learning from the United States’ DMCA safe harbor provisions. However, due to the different legal system, the attribution rules of indirect copyright liability in the United States have been ignored by China. The United States’ indirect copyright liability doctrines were developed through precedent; . accordingly, there have been no codified rules on the attribution of indirect liability for China to borrow. Thus China has stepped into a dilemma: on the one hand, the safe harbor rules play important roles in both attribution and exemption of liabilities; on the other hand, the 2010 Tort Liability Law and 2013 Provision also seek to serve as liability attribution rules. Since these rules have not been systematically organized and interpreted, a thorough reading of United States’ doctrines is helpful.

A. Fault-Based Liability Attribution Rules Developed from Common Law Cases

As a pioneer in the development of copyright laws based on the “early, rapid and widespread development of computer-based commerce”,[115] the United States occupies a dominant position in the development of indirect copyright liability laws which has impacted many other jurisdictions. Indirect (or secondary) liability[116] of service providers for online copyright infringement continues to be a highly controversial issue in the United States. Since Congress enacted its first copyright law in 1790[117], copyright law has been in constant conflict with evolving technological trends because technology reduces the ability of copyright holders to control their property.[118] This significant judicial dilemma led to the development of three indirect liabilities in the United States copyright regime: contributory liability and vicarious liability, which have been applied in a long line of cases, and inducement liability, which developed only in the last ten years.[119]

1. Is the Sony Rule Still Applicable in a Digital World? The Implied Factors in Contributory Liability

Contributory infringement in the United States is a common law liability regime. A defendant who engages in “personal conduct that encourages or assists the infringement” may be held liable for contributory infringement.[120] Courts began recognizing claims of contributory liability for infringment in the late nineteeth century, but only on the basis of intentional acts.[121] In one of the earliest cases involving contributory copyright infringement, Harper v. Shoppell,[122] the defendant was held liable as a joint tortfeasor for selling a printing plate, knowing that the purchaser would use it to make infringing copies. Later, in Kalem Co. v. Harper Bros,[123] the owner of a motion picture was held liable for authorizing the infringing activity by selling copies to exhibitors and supplying advertisements for the infringing exhibitions based on his knowledge that the illegal use would constitute an infringement. These early cases reflect the knowledge requirement to impose copyright liability. Gershwin Publishing Corp. v. Columbia Artists Management[124] developed a two-prong test for contributory infringement: “one who, with knowledge of the infringing activity, induces, causes or materially contributes to the infringing conduct of another, may be held liable as a ‘contributory’ infringer.”[125] In the 1980s, the Supreme Court again confronted the issue of contributory infringement in Sony Corp. v. Universal City Studios, Inc. (Betamax), which has been considered a conundrum in balancing conflicting interests between copyright holders and technology innovators.[126] In Sony, the District Court borrowed the “staple article of commerce” doctrine from patent law,[127] expressing the concern that commerce might be hampered if a mere constructive knowledge of possible infringement rendered the product distributor liable.[128] After a lengthy review of Constitutional provisions and case law.[129] the Supreme Court also recognized the role that the staple article of commerce doctrine played in balancing the interests of copyright holders and others’ freedom of commerce, holding that “the sale of copying equipment, like the sale of other articles of commerce, does not constitute contributory infringement if the product is widely used for legitimate, unobjectionable purposes. Indeed, it need merely be capable of substantial noninfringing uses.”[130] The Sony doctrine was tested in the landmark case of A&M Records, Inc. v. Napster, Inc. (“Napster”),[131] in which the Ninth Circuit refused to apply the Sony “staple article of commerce” doctrine to Napster for two reasons: (i) Napster had both the continuous ability to limit copyright infringement in ways that VCR manufacturers did not;[132] and (ii) actual knowledge of direct infringement rendered the Sony rule inapplicable.[133] The application of the Sony doctrine in Napster revealed that, in future cases with respect to mass-market means of copying, courts should “inquire into non-infringing uses when the distributor of the device lacks actual knowledge of and control over specific infringements.”[134] In addition, when adjudicating a case involving a dual-purpose product (one capable of substantial non-infringing uses), factors such as actual knowledge of the direct infringement and the ability to control direct infringement should also be taken into account. Sony was further tested in Aimster,[135] in which the Seventh Circuit disagreed with the interpretation of the Sony Rule by the district court and with Napster’s approach in application of the Sony Rule on the element of control.[136] Judge Posner reasoned from an economic perspective, recognizing that although control is a factor to be considered in determining contributory infringement,[137] the preliminary injunction the district court granted to the recording industry based on Aimster’s ability to control its users could result in the shutting down of the Aimster service, contrary to the clear meaning of the Sony decision.[138] The biggest challenge that the Sony rule and the doctrine of contributory liability encountered was the Grokster case, in which Grokster was sued by MGM for distributing free software products that allowed computer users to share copyrighted works through de-centralized peer-to-peer networks.[139] According to Grokster, under the Sony rule, the software it distributed was capable of substantial non-infringing uses.[140] Additionally, the remaining areas of uncertainty regarding the Sony “staple article of commerce” doctrine became the core of the case. Unlike Napster, which had actual knowledge of specific infringement and the ability to control direct infringement by operating a centralized system,[141] Grokster’s decentralized structure disclaimed its ability to obtain actual knowledge and prevent infringements once the product was distributed.[142] Without the two key elements established in the Napster decision, it seemed that Grokster was able to shield itself with the Sony defense. However, the Supreme Court disagreed on multiple grounds. Major disputes arose as to the interpretation and application of the Sony doctrine. Justice Souter refused to visit Sony further, but employed an inducement rule, holding that Sony did not apply when clear intent to infringe was demonstrated.[143] Justice Ginsburg argued that the Sony rule still applied, but not in this case, which “differ[ed] markedly from Sony,”[144] and the Ninth Circuit needed to reconsider the meaning of Sony.[145] Justice Breyer insisted on maintaining Sony, arguing that neither should it be strictly interpreted, nor should it be modified.[146] There have been a number of criticisms regarding the Sony “staple article of commerce” doctrine. The first is that the transplant of the Sony doctrine from patent law is a diversion from general tort law principles.[147] Some argue that the Supreme Court abused its discretion by extending its analysis of contributory and vicarious liability when the finding of fair use did not stop their analysis, which is possibly “out of context with the tradition of the Court not to engage in rule making beyond the case before it.”[148] Furthermore, articulating a standard from the Patent Act was “unfortunate and inapposite” for the readiness of technology assessment.[149] It was observed that two major problems arose with the application of the “staple article of commerce” equation in the digital world. First, there were no objective standards for the construction of the doctrine that would “prevent an inference of intent to result in infringement by the end user”.[150] Second, safe harbors provided the same function.[151] Accordingly, the objective of the “staple article of commerce” doctrine was misread. The purpose of the doctrine was for the protection of technology, and for ensuring that lawful use of a patent is not excluded, rather than “to create an inference of intent to infringe”.[152] The second criticism concerns the vague meanings of “capable of” and “substantial”. There has been debate as to whether “capable of” simply means current use of technology or also includes potential uses.[153] After all, “only the most unimaginative manufacturer would be unable to demonstrate that an image-duplicating product is ‘capable’ of substantial non-infringing uses.”[154] It is also evident, as some suggested, that eBay, Amazon, or Google and the multiple of other Internet and Information technologies be categorized as staples of commerce.[155] As for the meaning of “substantial”, it is arguable whether a minority non-infringing use would nonetheless be “substantial”.[156] Like Grokster suggested, if 10% of non-infringing uses in Sony were regarded as “substantial”, how about the 9% non-infringing uses in Grokster?[157] The questions such as how to identify the doctrine, what constitutes non-infringing use, and the time frame of measurement, are left open, which create fears of uncertain futures for innovators of new technologies.[158] Therefore, one may argue that a mature technology analysis is needed in order to conduct inquiries regarding the assessment of new technologies. Put another way, the problems raised by technology are better solved by technology itself. The third and the most important criticism is whether the Sony rule remains applicable in the digital world at all. Copyright law, the Court wrote, must “strike a balance between a copyright holder’s legitimate demand for effective . . . protection, and the rights of others to freely engage in substantially unrelated areas of commerce.”[159] That is why it was observed that “copyright law is important, but at some point copyright incentives must take a backseat to other societal interests, including an interest in promoting the development of new technologies and an interest in experimenting with new business opportunities and market structures.”[160] The driving concern in Sony was “a fear that indirect liability would have given copyright holders control over what was then a new and still-developing technology”.[161] That concern is why the Sony rule has played an important role in balancing the interest between copyright holders and technology innovators.[162] However, Grokster was substantially different from Sony in that the latter was used for individual non-commercial copying and the former was an unlimited copying tool without any restriction, which made mass-production possible. This difference demonstrates that the Sony rule, which was produced in the traditional dissemination world, is not suitable in the digital world anymore. More importantly, the vague and undefined wording in the Sony rule has triggered divergence in its interpretation. This was apparent in Grokster, where the Court had to apply inducing infringement in order to escape the difficulty of applying the Sony rule.[163] Creating a Sony rule might only be a temporary solution concerning the protection of a new technology. When challenged by new cases such as Grokster, it is time to consider whether the Sony rule is still applicable, and if not, what the alternatives are. Since the Grokster case, “ ‘inducement’ and ‘substantial non-infringing use’ will become legal conclusions, separating the Sony (good technology) sheep from the Grokster (evil entrepreneur) goats.”[164] While Robert I. Reis argued that “Sony left us with doctrine and dicta that obscured the need for rigorous methods of evaluation and assessment of new technologies that ensure reasonable standards and transparency”,[165] the interpretation of the Sony role has been evolving though cases, especially those with respect to new technologies. Since copyright holders started to target intermediaries such as ISPs as a shifting strategy under the digital environment, the scope of contributory copyright liability has been expanding with the development of technology with fear that technology would “unjustly enrich secondary actors at the expense of originators and destroy the latter’s creative incentives”.[166] The key factors to determine contributory liability are the defendant’s knowledge of direct infringement conducted by the third person and the material contribution to the infringement.[167] These two criteria have changed over time and continue to evolve, yet still remain “confusingly opaque” and not “suitably apportioned”,[168] especially under the challenge of P2P file sharing cases.[169] In order to determine the culpability of intermediaries, Dixon proposes a common set of elements to be considered by courts, including the relationship of the third party with the user, the extent of the third party’s involvement, knowledge of infringing activities, intention of the third party, extent of infringement and lawful activities, financial or other benefit of third party, ability to prevent or deter infringement, due care of third party and cost-benefit analysis.[170] Through presenting these elements and analyzing factors in different cases,[171] Dixon points out that “no one factor itself will impute liability, but the strong presence of two or more accumulated elements ties a third party more closely to the infringement in ways that courts may find sufficient to impose liability on the third party.”[172] The above elements were not expressly stressed by United States courts but have been taken into account in many occasions. For example, though the relationship between the direct infringer and the third party was not considered as an essential element of contributory liability in any United States court, it was brought about several times. In Sony, the district court noted that “Sony had no direct involvement with any Betamax purchasers who recorded copyrighted works off the air.”[173] In the dissenting opinion of the Supreme Court, Justice Blackmun argued that “the District Court reasoned that Sony had no direct involvement with individual Betamax users, did not participate in any off-the-air copying, and did not know that such copying was an infringement of the Studios’ copyright”, however, the Court stated that the contributory liability “may be imposed even when the defendant has no formal control over the infringer.”[174] In Napster, the court emphasized Napster’s ongoing relationship with its customers.[175] At any time, Napster could have refused service to users who were violating copyright law. VCR manufacturers, by contrast, had no such power; their relationship with any customer ended at the moment of sale. The element of control has not been officially recognized as a constituting element in deciding indirect copyright liability cases. However, it has functioned as an important nexus between primary and secondary infringers. The element of control is justified not only theoretically, based on fault, as a duty of care exists on the indirect infringer who has the ability to control, but also practically, based on the consideration for cost-efficient litigation.[176] In practice, courts have considered the factor of control not only in contributory infringement cases, through “the knowledgeable giving or withholding of a material contribution necessary to carrying out the infringing activity,” but also in vicarious liability cases, exercised “directly through the supervisory powers of the secondary over the subordinate primary”.[177] From the above analysis, it can be seen that the key factors analyzed in different liability forms should be interpreted under the specific circumstances. Other factors, such as relationship between direct and indirect infringers or due care of third party, are all invited into consideration.

2. The Expansion of Vicarious Liability in Copyright Law: Why It Does Not Apply in China?

Vicarious liability was developed out of the doctrine of respondeat superior, which provides that employers can be held strictly liable in general tort law for torts committed by their employees in the course of their employment.[178] The traditional formula states that:

When the right and ability to supervise coalesce with an obvious and direct financial interest in the exploitation of copyrighted materials –even in the absence of actual knowledge that the copyright monopoly is being impaired-the purposes of copyright law may be best effectuated by the imposition of liability upon the beneficiary of that exploitation.[179]

As a form of risk allocation,[180]the rationale for this form of liability is the incentive theory, which suggests that the employer is in a position to supervise and control the employee.[181]This liability was first extended[182]beyond an employer/employee relationship to cover the “dance hall” cases,[183]in which an independent contractor was found liable because the general contractor was in a better position to supervise and knew the identity of the subcontractor.[184]Gershwin extended the formula of vicarious liability in copyright case, in which a defendant “has the right and ability to supervise the infringing activity and also has a direct financial interest in such activities”.[185] In Napster, the district court determined that Napster had the right and ability to supervise its users’ conduct, because the evidence showed that Napster had the ability to block infringers’ access, retained the right to control access to its system, and had the ability to locate infringing material listed on its search indices as well as the right to terminate users’ access to the system.[186] This approach was challenged by Aimster, in which a cost-benefit analysis was conducted to prove that the meaning of control should not include the means to exclude the technology from the market.[187] Similarly, Grokster stated that one “infringes vicariously by profiting from direct infringement while declining to exercise a right to stop or limit it.”[188] Grokster further explained the “control” element of the vicarious liability test as the defendant’s “right and ability to supervise the direct infringer.”[189] Thus, under Grokster, a defendant exercises control over a direct infringer when he has both a legal right to stop or limit the directly infringing conduct, as well as the practical ability to do so.[190] In analyzing Perfect 10, the Circuit Court held the evidence did not support Google’s right and ability to limit the direct infringement of third-party websites.[191] Google’s ability to control was even weaker than Napster, which could terminate its users’ accounts and block their access to the Napster system.[192] On the element of direct financial interest, both the district court and the appellate court agreed that Napster had a direct financial interest in the infringing activity based on the finding that by attracting more users through the availability of protected works on its system, “Napster’s future revenue is directly dependent upon ‘increases in user base.’”[193] The tort law doctrine of vicarious liability, once applied in indirect copyright infringement cases, has expanded, with a broad interpretation of “control” and “financial benefit.” This liability approach would easily disturb many service providers that have certain ability to control their users activities and gain revenue based on advertisement or other business models. For China, vicarious liability does not apply for two reasons. First, there has been no basis or precedent in copyright law for vicarious liability, which is strictly restrained in the tort law regime of employer/employee relationship. Second, current development of vicarious liability in the United States shows that the interpretation of the constituting elements have not reached a stable status, which, if being transplanted to China’s copyright law, could easily become a problem.

3. Inducement Liability: What Could be Learned for China?

After the shutting down of Napster, Grokster made an architectural modification from Napster’s centralized file sharing function to a decentralized model, in order to not only avert actual knowledge, but also eliminate its ability to control. By dissatisfying both the knowledge and control elements, Grokster attempted to defeat both contributory and vicarious liability claims.[194] However, the Supreme Court borrowed an “inducement” theory of liability from patent law and held that:

One who distributes a device with the object of promoting its use to infringe copyright, as shown by clear expression or other affirmative steps taken to foster infringement, going beyond mere distribution with knowledge of third-party action, is liable for the resulting acts of infringement by third parties using the device, regardless of the device’s lawful uses.[195]

What conduct may be sufficient to render a defendant liable for inducement? The Supreme Court pointed out that the Sony rule limits imputing culpable intent rather than ignoring evidence of intent.[196] Thus, if evidence shows statements or actions directed to promoting infringement, the Sony rule will not preclude liability.[197] Grokster and Sony had different bases of liability for distributing a product open to alternative uses. Grokster emphasized the illegal objective from concrete evidence, while Sony stressed prohibiting the imputation of fault.[198] The classic case of direct evidence of unlawful purpose occurs when one induces commission of infringement by another, as by advertising. Under common law, one who “not only expected but invoked infringing use by advertisement” is liable for infringement “on principles recognized in every part of the law.”[199] According to the court, mental element or conduct alone is insufficient for a finding of indirect liability; an analysis must be based on all relevant factors.[200] Here, the summary judgment record was replete with other evidence that Grokster and StreamCast, unlike the manufacturer and distributor in Sony, acted with a purpose to cause copyright violations by illegal use of the software.[201] The Court looked to factors to determine inducement including “advertisement or solicitation that broadcasts a message designed to stimulate others to commit violations.”[202] To this end, the court declared that the respondents’“unlawful objective is unmistakable.”[203] Through importing the inducement rule from patent law, however, the Court failed to make further clarification on how far this rule should be stretched for future P2P illegal file sharing cases, especially on the question of “whether an inducer must have an intent to induce the acts that constitute infringement or an intent to induce infringement.”[204] Given the statutory structure in the patent law, a narrower standard was proposed, which required that the “inducer have an intent to induce infringement, not merely intent to induce acts that constitute infringement.”[205] This was a pro-competitive standard that encouraged newcomers to enter the market if they obtain a good faith belief in fair competition.[206] This approach solved the spiny Grokster case, and more importantly, cleared the path for the future application of indirect copyright liability rules on other P2P file sharing cases. China’s current legislation on online copyright infringement has been focusing on fault of a service provider, which was interpreted as actual or constructive knowledge.[207] However, this interpretation overlooked the culpable conduct of an abetting infringer. In its United States counterpart, the culpable conduct of an inducer is the purposeful, culpable expression and conduct, demonstrated by subjective and objective evidence. The ignorance of abetting infringement evidence in China’s legislation will cause trouble in deciding P2P file sharing cases when the P2P service providers have no knowledge and no control over the direct infringement. If elements of inducement infringement are to be imported to China, the service providers will not only avoid presenting unlawful objectives, but also take precautions and pay attention to the due care obligation such as implementing filter technologies.

B. Strict Interpretation of DMCA Safe Harbors

Designed as a balance between protecting the rights of copyright holders and promoting technology development, DMCA safe harbors have played a significant role in U.S. copyright law.[208] For technology developers, efficiency of internet services is assured and the variety and quality of the services continues to improve.[209] For copyright owners, an efficient remedy is provided without incurring substantial litigation fees.[210] There has been a significant amount of scholarly literature and judiciary opinions on the interpretation of the DMCA safe harbor rules, especially on the meaning of the knowledge and control requirements.[211] However, recent developments in American copyright litigation make it highly doubtful that the current interpretation is sufficiently clear. For one thing, Circuit Courts demonstrate split judiciary opinions on the content of knowledge, the specificity of the knowledge requirement, the qualifications of actual and red flag knowledge, the willful blindness doctrine, and the relationship between the knowledge requirement and the common law contributory infringement doctrine.[212] Similar questions arise when courts deal with the interpretation of the control requirement, e.g., whether item specific knowledge is required, whether the control requirement codifies vicarious liability, and what “something more” means in the application. These uncertainties have caused major confusion in litigations, which threatened to undermine the purpose of the DMCA safe harbors. In practice, after years of their application in litigations, the safe harbors have been under severe criticism as “a confusing and illogical patchwork” which “makes no sense;”[213] as very old;[214] and as baring deficiencies in vague and ambiguous language, which reflect the political compromise.[215] It is crucial to clear up the uncertainties in order to encourage both the protection of copyright and the development of new technologies.

1. Clarifying the Knowledge Requirement under § 512(c) and (d) of DMCA: Actual or Apparent Knowledge of Specific Infringement

Section 512(c)(1)(A) and section 512(d)(1) of the DMCA are similar, both providing that a service provider who stores “information residing on systems or networks at direction of users” or provides “information location tools” shall not be liable for monetary relief, if the service provider “does not have actual knowledge that the material or an activity using the material on the system or network is infringing”; and “in the absence of such actual knowledge, is not aware of facts or circumstances from which infringing activity is apparent”; or “upon obtaining such knowledge or awareness, acts expeditiously to remove, or disable access to, the material that is claimed to be infringing or to be the subject of infringing activity”.[216] “Knowledge” may refer either to knowledge of the activity itself or knowledge that the activity constitutes infringement. The latter argument is supported by an analysis of the statutory language, legislative history and case interpretation.[217] First, from the plain reading of the legislative language, it is evident that relevant provisions clearly require that the provider knows not only of the existence of the infringing material or activity, but also that it is infringing.[218] Second, with regard to the red flag test,[219] the committee reports reflecting legislative history made clear that “the red flag must signal to the provider not just that the activity is occurring, but that the activity is infringing”.[220] In other words, knowledge depends on whether the facts and circumstances make apparent the infringing nature of the user’s activity.[221] Case law has strengthened the above arguments. For instance, the Ninth Circuit in Perfect 10, Inc. v. CCBill LLC rejected the copyright owner’s allegation that the defendants must have been aware of apparent infringing activity because of the obvious nature of the domain names such as “illegal.net” and “stolencelebritypics.com”.[222] According to the court, the infringing nature could not be certain because the burden of determining whether photographs were illegal could not be placed on the service provider.[223] Thus, the relevant question was not “whether the defendants knew about the photographs, but whether they knew of the photos’ infringing nature”.[224] Another example is Corbis Corp. v. Amazon.com,[225] in which the court concluded that although Corbis sent Amazon notices, they did not constitute red flags, because Corbis was silent regarding the content of the complained listings, which meant Amazon had no clue of the infringing nature of those sales.[226] In this light, the court thus articulated the statutory “awareness” standard as more demanding than the common law “should have known” standard.[227] The second concern of the knowledge requirement of safe harbors is whether general or specific knowledge is required. This issue was dealt with in the lengthy Viacom v. YouTube case.[228] The court reached its finding from contextual analysis, structural analysis, legislative intent analysis, and case law analysis. First, the court read from the context of the statute.[229] In practice, service providers must expeditiously remove or disable access to the infringing material only if they can locate the material, which requires specific knowledge of the infringement.[230] Second, the structure and operation of the statute require the “specific knowledge” construction of the safe harbors. As pointed out by Judge Fisher in UMG Recordings, Inc. v. Shelter Capital Partners, LLC,[231] considerations of requiring specific knowledge of particular infringing activity were reflected in Congress’s decision to enact a notice and takedown protocol, and in the “exclusionary rule” that prohibited consideration of substantially deficient §512(c)(3)(A) notices which encourage the copyright holders to clearly identify specific infringing material,[232] rather than putting the monitoring obligation on service providers. This leads to the second consideration: the requirement of general knowledge would impose an obligation of policing infringement on service providers, which contradicts §512(m) of DMCA.[233] The view that requiring expeditious removal in the absence of specific knowledge or awareness would be “to mandate an amorphous obligation” and cannot be reconciled with the language of the statute.[234] Third, the legislative intent reflected in the Committee Reports also requires specificity of the knowledge. Both the Senate and House Reports noted that the Online Copyright Infringement Liability Limitation Act (OCILLA) was intended to provide “strong incentives for service providers and copyright owners to cooperate to detect and deal with copyright infringements.”[235] Copyright holders are better able to efficiently identify infringing copies than service providers “who cannot readily ascertain what material is copyrighted and what is not.”[236] The Reports also cleared the cloud in the construction of the actual and red flag knowledge by indicating that their difference is not between specific and general knowledge, but between a subjective and objective standard.[237] Case law also comports with the specific knowledge requirement. For instance, in UMG Recordings, Inc. v. Veoh Networks, Inc.[238], the district court concluded that “CCBill teaches that if investigation of ‘facts and circumstances’ is required to identify material as infringing, then those facts and circumstances are not ‘red flags’.”[239] The Court of Appeals reached the same conclusion by noting that “we do not place the burden of determining whether materials are actually illegal on a service provider.”[240] While the Ninth Circuit opinion in UMG III superseded UMG II, the argument regarding the specificity of knowledge remains the same: general knowledge is insufficient to meet both the actual and red flag knowledge requirement under §512 (c)(1)(A).[241] There have been two different approaches in interpreting the DMCA knowledge standard. One is the parallel or co-extensive approach that codifies the contributory infringement knowledge standard; the other is the independent approach that is different from, and more stringent than, the contributory infringement knowledge requirement. The copyright holder in litigation usually argues for the first approach, which sets up a lower standard of DMCA knowledge, under which the service provider will probably lose the benefits of the DMCA safe harbors. The service provider, on the contrary, argues for the latter approach, because a higher standard helps it become immune from contributory liability. In order to decide which approach should be applied, extensive consideration, including structural analysis of the legislation, the purpose of the legislators, and economic analysis is recommended, if possible. It was argued that the safe harbors and the common law contributory liability principle differ in their treatment of knowledge that comes by way of notice from a copyright owner. Contributory infringement recognizes notice as necessary knowledge for imposing secondary liability. However, safe harbor provisions provide a notice-and-takedown system that imposes more stringent requirements and differ from contributory infringement in operational detail. The form of notice decides whether the obligation of removing is triggered; thus, a noncompliant notice cannot create actual knowledge, according to the requirements of section 512. However, the common law contributory liability principle recognizes a noncompliant notice as sufficient to meet the knowledge requirement.[242] Regarding the “red flags” theory of liability, beyond the actual knowledge such as notice received from the copyright holder, under what circumstances must an ISP remove potentially infringing material in order to invoke the DMCA safe harbor? Under contributory liability, a defendant could be liable if he “knew or had reason to know of another’s direct infringement and materially contributed to it.”[243] However, under the DMCA safe harbor, possession of the knowledge will attract liability only if the ISP did not act “expeditiously to remove, or disable access to, the material.”[244] Is the level of knowledge between “should know” in contributory liability and “awareness” under the safe harbor equivalent? In order to determine the answer, Edward Lee conducted an analysis of DMCA text, structure, legislative history and case law.[245] First, from the plain language of the legislation, infringing activity must be “apparent”, which means it is “plain, clear, or obvious.”[246] This is a high standard of knowledge. Second, the structure of the DMCA safe harbors also supports adopting a high standard of awareness of “obvious” or “blatant” infringement, because a low standard would invite constant litigation against Internet companies and turn ISPs into censors, which Congress has expressly avoided.[247] Third, legislative history shows in the explanation by the Committee Report that the red flags are apparent from even a brief and casual viewing, with an important policy reason that the Congress “did not want to saddle Internet sites with the impossible burden of trying to differentiate what constitutes copyright infringement in the myriad of situations on the Internet.”[248] Fourth, all cases concerning the ISP safe harbors have applied a high standard of particularized knowledge, consistent with the above interpretation.[249] From the analysis above, section 512(c) and (d) of DMCA do not codify the common law principle of contributory liability for copyright infringement, because first, the knowledge requirement is different between statutory and common law principle in that the former requires not only knowledge itself but also knowledge of the infringing nature of the activity. Second, the common law and the statute treat notice differently in that the latter requires compliance in form. Third, the level of knowledge requirement is different between the common law contributory liability and the statute’s safe harbor provisions. Therefore, in interpreting the knowledge requirement in the DMCA safe harbors, an independent and narrow approach, rather than a paralleled and broad approach, is adopted.

2. “Control” and “Benefit” under § 512(c) and (d) of DMCA

The §512 safe harbor provides that an eligible service provider must “not receive a financial benefit directly attributable to the infringing activity, in a case in which the service provider has the right and ability to control such activity.”[250] What does “right and ability to control” exactly mean? Three questions arise as to the concept of control under the safe harbor provisions. First, is “item-specific” knowledge of infringement required in its interpretation of the “right and ability to control” infringing activity under 17 U.S.C. §512(c)(1)(B)? Second, does the control element in safe harbors codify the common law vicarious liability? Third, if not, what more does control exactly mean? As to the first question, in Viacom, the district court concluded that “the ‘right and ability to control’ the activity requires knowledge of it, which must be item-specific.”[251] In any event, the provider must know of the particular case before he can control it. If infringing material with sufficient particularity is identified as “red flags”, it must be taken down.”[252] However, on appeal, the court held that two competing constructions of the “right and ability to control” infringing activity were both fatally flawed.[253] The first construction that “the provider must know of the particular case before he can control it” was adopted by the district court in favor of the defendants.[254] The Ninth Circuit in UMG took a similar position that “until the service provider becomes aware of specific unauthorized material, it cannot exercise its “power or authority” over the specific infringing item.”[255] But the Second Circuit in Viacom held that the district court “erred by importing a specific knowledge requirement into the control and benefit provision,” and the case was remanded for further fact finding.[256] They disagreed on the aspect of literal interpretation of the statute and concluded that “importing a specific knowledge requirement into §512(c)(1)(B) renders the control provision duplicative of § 512(c)(1)(A).”[257] According to §512(c)(1)(A), a service provider that has specific knowledge of infringing material and fails to effect expeditious removal would be excluded from the safe harbor protection, and therefore the existence of §512(c)(1)(B) would be superfluous. The second construction of the “right and ability to control” was that it codifies the common law doctrine of vicarious copyright liability, evidenced by the House Report relating to a preliminary version of the DMCA:

The “right and ability to control” language . . . codifies the second element of vicarious liability . . . Subparagraph (B) is intended to preserve existing case law that examines all relevant aspects of the relationship between the primary and secondary infringer.[258]

However, this codification reference was omitted from the committee reports describing the final legislation. Before the district court on remand gave its decision in Viacom v. YouTube, the Ninth Circuit issued a decision on the same issue in UMG III, following the Viacom opinion by the Second Circuit, holding that there were several reasons for a stricter interpretation of the “right and ability to control” than vicarious liability, in light of the DMCA’s language, structure, purpose and legislative history. First, the term “vicarious liability” is mentioned nowhere in §512(c), and the language used in common law standard “is loose and has varied”.[259] Second, considering the structure of §512(c), if the ability to control is being read as the ability to remove or block access, “the prerequisite to §512(c) protection under§512(c)(1)(A)(iii) and (C), would at the same time be a disqualifier under§512(c)(1)(B) where the ‘financial benefit’ condition is met”, which means that a catch-22 is created by Congress.[260] Applying this interpretation would “defeat the purpose of the DMCA and render the statute internally inconsistent.”[261] Third, according to the legislative history, though it was not suggested to codify the element of control as vicarious infringement, this suggestion was omitted from later reports.[262] Fourth, Congress explicitly stated that “the DMCA was intended to protect qualifying service providers from liability for all monetary relief for direct, vicarious and contributory infringement.”[263] In addition, it was clear that “the Committee decided to leave current law in its evolving state and, instead, to create a series of ‘safe harbors,’ for certain common activities of service providers.”[264] Furthermore, if Congress had intended the control element be coextensive with vicarious liability law, “the statute could have accomplished that result in a more direct manner.”[265] According to above analysis, the “right and ability to control” infringing activity under §512(c)(1)(B) “requires something more than the ability to remove or block access to materials posted on a service providers [website].”[266] Courts tended to interpret the phrase “right and ability to control” as “exerting substantial influence on the activities of users, without necessarily—or even frequently—acquiring knowledge of specific infringing activity.”[267] Therefore, the case was remanded to the district court to consider whether YouTube had the right and ability to control the infringing activity and received a financial benefit directly attributable to that activity.[268] Since the Ninth Circuit held that the “right and ability of control” does not codify the common law vicarious liability and requires “something more” than “just ordinary power over what appears on the provider’s website,”[269] what constitutes “something more?” In UMG III, the Court addressed “high levels of control” and “purposeful conduct” as two standards of “substantial influence” that the service provider must exert on the activity of users.[270] In this case, the evidence presented was not enough to create the issue equivalent to the activities found to constitute substantial influence. Accordingly, the element of “right and ability to control” was not applied to Veoh Networks, which met all of the §512(c) requirements. In Viacom IV, the plaintiffs claimed that “something more” was established by both YouTube’s willingness and ultimate editorial judgment and control over infringing content. This was shown by YouTube’s decisions “to remove some but not all infringing material, by its efforts to organize and facilitate search of the videos appearing on the site, and by its enforcement of rules prohibiting, e.g., pornographic content.”[271] The court took a very strict approach in interpreting the evidence that alleged YouTube’s influence or participation in the infringing activity occuring on its site. The court found that “something more” required by the “right and ability to control” must only be fulfilled when the defendant exercises substantial participation or ultimate editorial judgment over the infringing activity.[272] Evidence provided by the plaintiffs demonstrated neither participation in, nor coercion of, user infringement activity. Therefore, YouTube did not have the right and ability to control infringing activity within the context of §512(c)(1)(B). As discussed, the question arises in Viacom v. YouTube as to whether the safe harbor provision in DMCA codifies the common law principle of vicarious liability for copyright infringement. The court’s answer was no, and some commentators agree.[273] The resemblance of factors such as “right and ability to control” and “direct financial interest/benefit” between vicarious liability and safe harbors leads to a loophole theory that the DMCA safe harbors provide no immunity from vicarious liability at all, because safe harbors and vicarious liability share the exact same standard.[274] Mark Lemley indicates that the language of DMCA safe harbors suggests that it provides a safe harbor under section 512(c) only against claims of direct and contributory infringement, rather than vicarious liability. However, the legislative history suggests the opposite. Thus, a digital hole is created.[275] However, Edward Lee provided the interpretation that the DMCA safe harbors provide partial immunity from some, but not all vicarious infringement claims. He explained that “the term “receive a financial benefit directly attributable to the infringing activity” requires a closer causal connection between the infringing activity and the ISP’s actual receipt of a financial benefit. It must be “directly attributable” to the infringing activity, which is a higher level of proof and causation than required under the common law”.[276]

III. Establishing Indirect Liability System for Digital Copyright Infringement in China

Current Chinese tort law has a logic gap that had existed long before the promulgation of the 2010 Tort Liability Law. Unfortunately, the new legislation has not yet recognized the concept of indirect liability, nor has it developed an independent basis for indirect infringement. Therefore, the tort law needs to be amended, adding an “overlapping tort” as one of the liability forms and theoretical bases for indirect liability. Constructing indirect liability forms and standards for online copyright infringement involves three steps. The first step is to establish a general rule for indirect copyright liability. This rule requires three key components: direct infringement as a prerequisite, because indirect infringement does not exist without direct infringement; a subjective fault including intent or knowledge; and enablement, that the indirect infringer provides means for direct infringement. The second step constructs two types of indirect liability. Contributory liability is mainly designed for ISPs that provide services such as hosting and information locating, while inducement liability can solve many problems with the issue of P2P infringement, especially for ISPs that provide decentralized software. The third step requires strict interpretation and application of safe harbor provisions. These liability limitation rules should not unduly impede legitimate digital communications, nor should they unreasonably influence the Internet, which has been an effective communications platform, commercial channel and educational tool. Specifically speaking, it is proposed that China’s legal framework for indirect copyright infringement consists of relevant articles in the 2010 Copyright Law, the 2010 Tort Liability Law and the 2013 Provision as liability attribution, and related rules in the 2006 Regulation work as copyright liability limitation. However, considering the ambiguities in current laws, I make the following recommendations. To begin with, the 2010 Copyright Law shall explicitly recognize indirect copyright liability. Unlike the United States, which has developed indirect copyright liability in case laws, judicial decisions in China have no binding effects.[277] In this light, it is essential for legislators to codify this theory and put it into the statute, as long as the theory matures in tort law as well. Next, China has transplanted the United States safe harbor provisions, which have served as both attribution and exemption of liabilities, owning to the misunderstanding of safe harbor provisions. This article suggests a strict interpretation of safe harbors, rather than considering safe harbors as a codification of common law principles, based on analysis of legislative history, legislators intent and context reading. Investigating into the nature, role and true meaning of the DMCA safe harbors resolves the confusion that exists in China’s legislation, such as the nature of safe harbor provisions, contradictable knowledge standards, and contested control and benefit requirements. I therefore propose a revision of current Chinese safe harbor provisions. The second paragraph of article 23[278] should be removed for two reasons: first, it is an expression of liability attribution rather than liability limitation. Since the 2006 Regulation shall serve as a safe harbor for NSPs, there is no need for the paragraph to exist. Second, it has been proved that NSPs assume independent liability based on the overlapping tort theory rather than joint liability. The existence of this article will create confusion in the future application of law. The other proposition for the modification of the 2006 Regulation is that the article 22(4)[279] should be removed. As demonstrated previously, vicarious liability lacks root in Chinese copyright law system, and the stipulation on financial benefit as one limitation to the liability renders a higher copyright liability for NSPs than that in the United States. This is disproportionate for NSPs in China. Lastly, concern has been raised towards the culpability of NSPs, especially the inconsistency of the knowledge standard. It is vital for both courts and scholars to reach the consensus as to the interpretation and application of the knowledge requirement such as “know,” “should have known” and “have reasonable ground to know.” Great achievement has been made in United States on the theory of knowledge, such as the content of knowledge, the generality of knowledge and the meaning of “red flag” knowledge. China can benefit from the United States experience. In addition, China’s laws have been partially emphasizing the mental element of the defendant, but have overlooked the objective aspects such as culpable conduct.

Conclusion

Ever-improving technological advances, especially the development of the Internet and digital technology, have provided impetuses as well as challenges for the application of traditional copyright law principles, which have been tested within the digital environment to see whether the balance between copyright and commerce has been interrupted and whether changes are needed. The indirect copyright liability regime becomes the crux of the contradiction between copyright law and technology by imposing liability on intermediaries who do not commit copyright infringement directly but are held liable for infringement committed by others, based on efficiency and moral grounds. Within the digital environment, ISPs are facing potential liability for the acts of subscribers who are using their services to access, upload or download information. However, a lack of certainty of ISP liability in current digital legislation will inevitably decrease ISPs’ incentives for utilizing new technology and participating in e-commerce. Consequently, this may have a negative impact on the progress of science and on economic development in China. China has been importing the United States’ safe harbor models into its own legislation, which, however, has caused confusion from two aspects. First, China and the United States have different legal systems, which makes the transplant inapplicable in certain ways. Second, unlike the United States’ safe harbor rules that supplement the attribution of indirect liability principles, such as contributory and vicarious infringements developed in case law, China has not fully developed its own attribution of liability principles, just like water without a source and a tree without roots. This study finds that in the United States, judicial practice of applying the doctrine of contributory infringement liability has abrogated the Sony rule by applying relevant factors, which have displaced the “substantial non-infringing uses” standard. Additionally, courts have been considering all relevant factors—including the relationship, control, means, knowledge, due care and inevitability of infringement, allowing for a wide spectrum of evaluation that work in a correlative pattern. Further, this study argues for a strict interpretation which considers integrity, clarity and uniformity as guiding principles for the efficient application of safe harbors. With a deep understanding of the United States’ approach in constructing the indirect copyright liability doctrines and safe harbors, a suitable approach for China could be found. In conclusion, this article proposes to establish a tort law-oriented, culpable conduct-based indirect copyright liability system, with modified safe harbor provisions, in China. In this way, a justified and compatible indirect liability system can be optimized with equilibrium among relevant parties.

---