When Knowing Is Not the Same as Intending
This series examines Cox Communications, Inc. v. Sony Music Entertainment, the Supreme Court's 2026 decision about when an Internet service provider can be held responsible for copyright infringement committed by its users. The immediate focus is narrower than the general debate over online piracy: does an ISP's knowledge that a connection has been used for infringement, combined with its decision to keep providing service, prove that the ISP intended to help the infringement? The answer matters because general-purpose networks carry enormous amounts of lawful activity at the same time that they can be used for unlawful copying.
The case began with a billion-dollar jury verdict. Sony Music and other copyright owners argued that Cox Communications contributed to infringement by continuing to serve subscribers whose Internet connections had been associated with repeated notices. A monitoring company sent Cox 163,148 notices over roughly two years. The Fourth Circuit accepted a rule under which supplying a product while knowing that the recipient would use it to infringe could be enough for contributory liability. The Supreme Court reversed.
The reversal did not say that copyright owners must tolerate services designed to promote piracy. Nor did it decide every form of liability that can reach an intermediary. It addressed contributory liability, a form of secondary liability: liability for someone else's infringement. The Court said that the provider must have intended its service to be used for infringement. On the majority's account, that intent can be shown through active inducement or through a service tailored to infringement. Cox's ordinary Internet access, the Court concluded, fit neither description.
The key distinction is easy to state but difficult to preserve in a real dispute. Knowledge answers one question: what did the provider know about the use of its service? Intent answers another: what did the provider mean for its service to help accomplish? A third question concerns inaction: what did the provider fail to do after learning about the conduct? The Supreme Court refused to treat those questions as interchangeable. This series follows that separation from the facts of Cox through the governing precedents, the DMCA safe-harbor system, and the disagreement exposed by Justice Sotomayor's concurrence.
The Case Behind the Slogan
Sony did not sue only the people who uploaded or downloaded copyrighted music. It sued Cox, the company that supplied the Internet connections those users employed. That choice put an intermediary problem at the center of the case. The Copyright Act directly makes a person who violates a copyright owner's exclusive rights an infringer. But the Act does not expressly say that every provider of the means of infringement is liable for another person's acts. Courts have developed forms of secondary liability, including contributory and vicarious liability, through precedent and common-law principles.
Only contributory liability was before the Supreme Court. Sony's separate vicarious-liability theory had prevailed at trial but was reversed by the Fourth Circuit, and the Supreme Court did not grant review on that issue. This procedural boundary is important. The decision is not a declaration that an ISP can never face copyright liability. It is a decision about whether the particular theory of contributory liability could rest on Cox's continued service to connections associated with infringement.
Cox served about six million subscribers. Each account was associated with an Internet Protocol address, commonly called an IP address. That association gave Cox useful information: it could generally connect an IP address to a subscriber account. But an address does not necessarily identify the person who performed a particular act. A household, coffee shop, dormitory, hospital, or university network may put many users behind one connection. A regional ISP may buy service from a larger provider and then pass it to thousands of customers. The provider can know that activity occurred through a connection while lacking a way to identify the individual user who performed it.
Sony's notices came from MarkMonitor, whose software detected apparently infringing uploads or downloads and traced them to IP addresses. That evidence was not trivial. It could show a recurring pattern associated with a particular connection. But it also had a built-in limit: a connection-level notice is not automatically person-level knowledge, and a recurring notice is not automatically proof that the network provider wanted the infringement to succeed.
Cox said it did not simply ignore the notices. After a second notice for an account, it sent a warning. Further notices could lead to a temporary interruption, a phone warning, or a suspension. After 13 notices, the subscriber could face termination of Internet service. Cox's contract also prohibited subscribers from using the connection to post, copy, transmit, or disseminate infringing material. Sony presented a different picture, emphasizing that Cox terminated only 32 subscribers for infringement during the relevant period and pointing to employee statements about protecting subscriber revenue. Cox responded that its warnings and suspensions ended 98 percent of the identified infringement. Those competing facts show why a litigation record cannot be reduced to a single number.
The Legal Question Is Narrower Than “Did the ISP Know?”
The Supreme Court's reasoning rests on two established examples. First, a provider may induce infringement by taking active steps to encourage it. The Court pointed to MGM v. Grokster, where file-sharing companies promoted their software as a tool for infringement and built their business around that use. Advertising, instructions, or other specific conduct can reveal an affirmative objective even when a product also has lawful uses.
Second, a provider may supply a service tailored to infringement. The Court drew this principle from Sony Corp. of America v. Universal City Studios, the Betamax case. The video recorder could be used to make unauthorized copies, but it also had substantial lawful uses, such as recording a television program for later personal viewing. Because the product was capable of substantial or commercially significant noninfringing uses, selling it to the general public did not by itself establish contributory infringement.
These examples create a more demanding question than whether infringement was foreseeable. What does the provider's conduct show about the service's purpose? Does the provider promote infringement? Does it shape the service so that infringement is its dominant or specially enabled use? Or does it provide a broadly useful tool that some users misuse? Lawful capability alone does not excuse active promotion, as Grokster demonstrates. But the possibility, or even the known occurrence, of unlawful use does not by itself transform a general-purpose service into a piracy device.
On the majority's reading, Cox did not induce its subscribers to infringe. Sony offered no evidence that Cox marketed Internet access as a way to obtain copyrighted music or instructed users how to do so. Cox's warnings and suspensions pointed in the opposite direction. Nor was Cox's service tailored to infringement. Cox supplied Internet access, which people use for communication, education, work, commerce, entertainment, and countless other lawful purposes. The service did not become infringement-focused merely because some subscribers used it to share music.
That is why the Court rejected the Fourth Circuit's formulation. The lower court had treated “supplying a product with knowledge that the recipient will use it to infringe” as sufficient culpable conduct. The Supreme Court said that approach created a new category beyond the two forms recognized in its precedents. Its repeated warning was that contributory liability cannot rest only on knowledge of infringement plus insufficient action to prevent it.
The result should not be summarized as “knowledge never matters.” Knowledge can matter under a different legal rule, in a safe-harbor analysis, or as part of a larger body of evidence showing purposeful assistance. The narrower point is that knowledge does not complete the intent element for this contributory-liability theory. A provider's awareness, its response policy, and the design or promotion of its service must be analyzed as distinct facts rather than compressed into the label “known infringer.”
The Questions This Series Keeps Separate
The DMCA adds another layer to the analysis. Section 512 creates safe harbors—limitations on liability—for providers that perform qualifying functions and satisfy specified conditions. A mere-conduit ISP may fall under section 512(a) for transmitting, routing, or providing connections, while hosting, caching, and linking services have different safe-harbor categories and different operational requirements. The safe harbor is a defense or limitation; it is not an independent rule saying that serving a known infringer is itself infringement.
That distinction explains why the Supreme Court rejected Sony's argument that its interpretation would make the safe harbor meaningless. The Court emphasized that section 512(l) says failure to qualify for a safe harbor does not count against a separate defense that the provider's conduct is not infringing. Cox could not rely on the safe-harbor defense for the relevant period because of an earlier ruling, but that did not prove the underlying contributory infringement. A provider can lose a defense without losing every argument on the merits, just as a provider can avoid one liability theory without earning a blanket immunity.
The DMCA's conditions still matter. Section 512(i) addresses repeat-infringer policies, and the Copyright Office has described those policies as part of Congress's attempt to exchange legal certainty for cooperation against online infringement. The Office also describes disputes over actual knowledge, red-flag knowledge, willful blindness, notice-and-takedown, and the absence of a general duty to monitor. Those questions are especially important for hosting and information-location services, whose functions differ from the mere provision of a connection. They should not be silently folded into the Supreme Court's narrower contributory-intent holding.
The central discipline for reading Cox v. Sony is therefore to ask which question is being answered. What did the provider know? What could it identify? What did it do after receiving information? What function did its service perform? Did it promote infringement or tailor the service to it? Is the argument about substantive liability, or about eligibility for a statutory safe harbor? The majority's answer is clear on the case before it: a general-purpose ISP's knowledge of infringement and continued provision of access, without inducement or an infringement-tailored service, did not establish contributory liability.
The next unresolved issue follows directly from that answer. If the connection is not the person, and knowledge is not intent, what additional evidence would justify moving from ordinary service to purposeful facilitation? The answer begins with the network facts rather than with a slogan about “known infringers.”