For technology companies, platforms and businesses developing AI tools, Cox Communications, Inc. v. Sony Music Entertainment, 607 U.S. __ (2026), offers an important lesson about how a company communicates the intended use of its product.
The case involved Cox internet subscribers who used peer-to-peer networks to share copyrighted music without permission. Sony Music Entertainment and other copyright owners sent Cox more than 163,000 notices identifying IP addresses associated with suspected infringement. They argued that Cox knew repeat infringement was occurring but continued providing internet access because it wanted to retain subscriber revenue.
The Supreme Court held that this knowledge was not enough to make Cox contributorily liable. Sony also needed to show that Cox intended for its service to be used for infringement, either by encouraging unlawful conduct or supplying a service tailored to it.
Cox’s internet service had extensive lawful uses, and the company did not promote illegal file sharing. It also expressly prohibited subscribers from using their connections to infringe copyrights. Although that policy did not resolve the case by itself, it helped establish that infringement was contrary to the purpose Cox communicated for its service.
For businesses, the practical lesson begins there. Companies should state clearly that their products and services are intended for lawful use and that infringement is prohibited. Those expectations should appear in terms of service, user policies, marketing, onboarding materials and communications with customers.
The more difficult issue is whether the rest of the business supports that message. A company cannot rely on a written prohibition if its product features, marketing or growth strategy encourage the conduct its policies reject. After Cox, courts may look beyond what a company says to examine what it built, promoted and rewarded.
At Minx Law, the decision points to a broader shift: product architecture is becoming part of the legal record. The most important evidence in the next platform copyright case may not be an infringement notice. It may be the complete story the company tells through its policies, product decisions and commercial strategy.
What the Supreme Court decided
A jury originally found Cox contributorily and vicariously liable for infringement of 10,017 copyrighted works and awarded $1 billion in statutory damages. The Fourth Circuit later reversed the vicarious-liability ruling but upheld contributory liability, reasoning that Cox continued providing internet service to subscribers it knew would use the service to infringe.
The Supreme Court rejected that approach because it treated knowledge as though it were intent.
Writing for seven Justices, Justice Thomas explained that a service provider may be contributorily liable when it actively induces infringement. In MGM Studios Inc. v. Grokster, 545 U.S. 913 (2005), for example, file-sharing companies promoted their software as a way to obtain copyrighted works and developed their business models around that use.
Liability may also arise when a company provides a service tailored to infringement because the service lacks substantial or commercially meaningful lawful uses. This principle draws from Sony Corp. of America v. Universal City Studios, 464 U.S. 417 (1984), where the Court declined to hold the manufacturer of the Betamax recorder liable because the technology was capable of legitimate uses.
Cox did not fit either category. Its service was ordinary internet access, and the company did not market or design that service as a tool for illegal file sharing. Cox may have known that some subscribers were infringing and may have had financial reasons to retain them, but neither fact established that Cox wanted the infringement to occur.
This distinction is important for executives because knowledge, financial benefit and intent are different legal concepts. A company may know that customers are misusing a product and earn revenue from those customers without necessarily intending to support the misconduct.
That does not mean the company should ignore what is happening. It means that the legal analysis will consider the company’s own conduct rather than treating user misconduct as proof of the company’s purpose.
Justice Sotomayor, joined by Justice Jackson, agreed that Cox was not liable on the evidence presented but disagreed with the majority’s narrow definition of contributory liability. She argued that traditional aiding-and-abetting principles could provide another path when a company knowingly and intentionally assists infringement.
All nine Justices therefore agreed on the result, but they did not fully agree on how courts should analyze future cases. That disagreement leaves room for further development, particularly when a provider has more information about its users or more control over how its service is used.
Policies should reflect the company’s real position
The immediate business response to Cox should be straightforward. Companies that offer products capable of misuse should clearly explain their expectations for lawful use.
Terms of service should prohibit infringement in language appropriate to the product. User policies should explain which activities are restricted. Onboarding materials and customer communications should reinforce those expectations, particularly when a service allows users to upload, generate, distribute or monetize content.
Marketing deserves the same attention. A company should not promise responsible use in its legal terms while promoting capabilities that invite customers to reproduce protected work without permission. The inconsistency may become more important than either statement viewed alone.
Clear policies are not only defensive documents. They help establish the purpose for which the company offers its service and create a basis for responding when users depart from that purpose. They can also guide product, sales and customer-service teams when questions arise.
The policy must still be credible. A prohibition that is never communicated, monitored or enforced may provide a weaker account of the company’s intentions than one reflected throughout the business.
Companies do not need to prevent every possible infringement. General-purpose products routinely have lawful and unlawful uses, and the Supreme Court’s decision protects businesses from being treated as infringers merely because misuse is foreseeable.
They should, however, be able to explain the difference between an unwanted misuse of the product and a use the company has chosen to encourage. That explanation becomes harder when business decisions repeatedly favor the prohibited activity.
Product choices may reveal more than legal terms
Companies rarely express unlawful intent directly. The more difficult cases will examine whether intent can be inferred from a series of product and commercial decisions.
A platform may prohibit infringement while developing tools requested primarily by users seeking unauthorized content. An online marketplace may maintain a takedown process while allowing search and recommendation systems to direct customers toward obvious copies. An AI company may publish responsible-use terms while marketing its ability to reproduce recognizable creative work.
No single fact necessarily establishes contributory liability. Taken together, those choices may help a court understand whether infringement remained an unwanted use or became part of the product’s appeal.
This is where the Cox decision becomes more significant than its facts. The case concerned a general internet provider with limited information about the individuals using each connection. Many modern platforms have a closer relationship with their users and much greater control over what their products can do.
They collect customer data, recommend content, design creation tools and decide which capabilities to emphasize. They may also have detailed records showing why a feature was developed and which users the company expected it to attract.
For those businesses, product governance and copyright strategy are becoming difficult to separate. Decisions about safeguards, feature requests and customer targeting may later help explain what the company intended its service to support.
Legal review should therefore begin before a product launches or a new capability is introduced. Counsel should understand what the feature does, why customers want it, how it will be described and whether its meaningful value depends on access to protected material.
This is not about asking lawyers to design products. It is about recognizing when a product decision also communicates a legal position.
AI will test the limits of Cox
Generative AI presents an especially difficult test because AI companies often exercise more control over their systems than Cox exercised over its subscribers.
Cox provided a general internet connection. It did not create the infringing music files, determine what subscribers downloaded or always know which person was using a particular account.
An AI company may decide how a model is trained, what capabilities it offers, which requests are restricted and how closely the system may reproduce existing work. Some AI cases may also involve allegations of direct infringement by the company itself, which would raise issues beyond the contributory-liability claim addressed in Cox.
Technology providers will nevertheless rely on Cox when their products have substantial lawful uses. They will argue that knowledge of some infringing outputs does not establish that the company intended to produce them.
The strength of that position may depend on whether the company’s policies and product history support the same story. Courts may examine whether a disputed feature was created for lawful expression or to satisfy demand for imitation, whether marketing highlighted the ability to reproduce protected material and whether safeguards were weakened to attract users seeking those results.
These questions are already being answered inside companies, even when no one describes them as copyright questions. Product teams decide how closely a system may imitate existing work. Marketing teams decide which examples will generate attention. Leadership decides whether a controversial capability creates acceptable risk.
By the time litigation begins, those decisions may form the most important evidence of intent.
A more complete approach to copyright risk
Cox gives businesses offering legitimate, general-purpose services meaningful protection. It confirms that a company does not become contributorily liable merely because it knows some users are infringing or fails to prevent every violation.
The decision should not be reduced to permission to do less. It gives companies a reason to communicate their lawful purpose clearly and ensure that the rest of the business remains consistent with it.
Executives should be able to answer several practical questions: Do the company’s terms expressly prohibit infringement? Are those expectations communicated to users? Does the product encourage a use the policy rejects? Can the business explain the lawful value of a disputed feature? Do marketing and growth decisions support the company’s stated position?
For copyright owners, the enforcement inquiry must also become more complete. Notices remain important, but they may be the beginning of the investigation rather than its strongest evidence. Understanding what an intermediary built, how it promoted the service and which uses it cultivated may be essential to proving intent.
The intent gap makes some claims harder, but it also brings greater attention to the company’s own decisions. Businesses that want the protection available to neutral technologies should ensure that neutrality is expressed in their policies and reflected in their conduct.
After Cox, a company’s product may become its most important witness.
