After Cox v. Sony: The Supreme Court Rebuilds the Wall Around ISP Safe Harbors

In a unanimous March 2026 judgment, the Supreme Court reversed the $1 billion verdict against Cox Communications, holding that knowledge alone cannot make an internet provider a contributory infringer — reshaping how the DMCA's § 512 safe harbor matters.

In Cox Communications, Inc. v. Sony Music Entertainment, No. 24-171, 607 U.S. ___ (2026), decided March 25, 2026, the Supreme Court of the United States unanimously reversed the judgment of the U.S. Court of Appeals for the Fourth Circuit and remanded, holding that an internet service provider cannot be held contributorily liable for copyright infringement merely because it continues to serve subscribers it knows are infringing. Justice Thomas wrote for a seven-Justice majority (joined by the Chief Justice and Justices Alito, Kagan, Gorsuch, Kavanaugh, and Barrett); Justice Sotomayor, joined by Justice Jackson, concurred only in the judgment, sharply disputing the majority’s reasoning. The decision caps litigation that began with a 2019 jury verdict in the Eastern District of Virginia and a Fourth Circuit opinion at 93 F.4th 222 (4th Cir. Feb. 20, 2024) (No. 21-1168).

At a glance

Holding: Knowledge that a service “will be used by some to infringe” is not enough for contributory liability. Liability attaches only where the provider intends the infringing use — provable in two ways: (1) inducement (active encouragement of infringement) or (2) supplying a service “not capable of substantial or commercially significant noninfringing uses.”
Disposition: Fourth Circuit reversed; case remanded. The Fourth Circuit had already vacated the original ~$1 billion award (for more than 10,000 musical works) and rejected vicarious liability; the Supreme Court did not take up vicarious liability.
Why § 512 matters here: Cox had forfeited the DMCA § 512(a) safe harbor for the relevant period because the trial court found it failed to reasonably implement a repeat-infringer termination policy. That forfeiture is what exposed Cox to the secondary-liability theories the Court has now narrowed.
The split that wasn’t unanimous: All nine Justices agreed Cox should win, but Justices Sotomayor and Jackson would not have reworked the doctrine of contributory liability to get there.

The § 512 backdrop: losing the harbor is no longer the end of the story

The DMCA’s safe harbors, 17 U.S.C. § 512, are conditional. A conduit provider like Cox qualifies for § 512(a) immunity only if, as a threshold matter under § 512(i), it has “adopted and reasonably implemented” a policy providing for termination of subscribers who are repeat infringers. In the run-up to this case, the trial court concluded Cox could not satisfy that condition — its termination practices were treated internally as a revenue-protection exercise rather than a genuine enforcement mechanism — and so the safe harbor was foreclosed.

For years, the practical lesson plaintiffs drew was straightforward: prove a provider’s repeat-infringer policy is a sham, strip away § 512, and the door to massive secondary-liability damages swings open. That is precisely the path that produced the billion-dollar verdict. The Court’s opinion does not disturb the § 512(i) requirement itself. What it disturbs is the assumption on the other side of the safe-harbor wall. Justice Thomas reframed the DMCA as what it textually is — a set of defenses, a shield rather than a sword. The statute “does not expressly impose liability” on a provider that serves known infringers; it merely immunizes compliant providers from liability that must be established independently under ordinary copyright principles. Losing the safe harbor, in other words, returns a defendant to the baseline of common-law secondary liability — and that baseline, the Court held, requires intent, not mere knowledge.

The new intent standard for contributory infringement

The heart of the opinion is a tightened test. Drawing on Sony Corp. of America v. Universal City Studios (1984) and Metro-Goldwyn-Mayer Studios Inc. v. Grokster, Ltd. (2005), the majority held that a company “is not liable as a copyright infringer for merely providing a service to the general public with knowledge that it will be used by some to infringe copyrights.” Contributory liability turns on whether the defendant intended its service to be used for infringement, and that intent may be shown in only two ways: by inducing infringement through affirmative, specific acts of encouragement, or by offering a service not capable of substantial noninfringing uses.

A general-purpose broadband network plainly clears the Sony “capable of substantial noninfringing uses” bar — the internet is the paradigmatic dual-use technology. And the record contained no inducement: Cox did not market its service as a piracy tool or encourage subscribers to infringe. The Fourth Circuit had nonetheless sustained liability on a theory that Cox “knew of specific instances” of repeat infringement, traced them to identifiable subscribers, and kept those subscribers connected to protect monthly revenue. The Supreme Court held that this culpable-knowledge-plus-continued-service theory is not enough. Knowledge, even willful blindness to specific known infringers, “is not enough” absent inducement or a non-substantially-noninfringing service.

Because willfulness — and thus the enhanced statutory damages that drove the award toward $1 billion — was entirely contingent on the underlying contributory finding, the reversal swept the damages predicate away with it.

The Sotomayor concurrence: a warning about the safe-harbor bargain

Justice Sotomayor’s separate opinion, joined by Justice Jackson, agreed Cox should prevail on this record but accused the majority of an “artificial limiting of secondary liability” that is “supported by neither precedent nor statute.” Her institutional concern is pointed and directly implicates § 512: Congress built the DMCA safe harbors on the assumption that providers face real exposure to secondary liability if they do not police repeat infringers. Strip the underlying liability down to inducement and tailored-to-infringe services, and the elaborate § 512(i) compliance regime risks becoming an incentive structure with little behind it. If a provider faces almost no liability whether or not it terminates repeat infringers, the statutory carrot loses much of its bite. The majority answered that policy is for Congress, and that the DMCA’s text creates defenses, not affirmative duties enforceable through expansive common-law liability.

A doctrine now in tension with the courts of appeals

The decision lands directly on a developing body of appellate law. In UMG Recordings, Inc. v. Grande Communications Networks, LLC, No. 23-50162 (5th Cir. Oct. 9, 2024), the Fifth Circuit affirmed a contributory-infringement verdict against another conduit ISP on reasoning materially similar to the Fourth Circuit’s — knowledge of specific repeat infringers plus continued service and material contribution — while vacating the $47 million damages award on an unrelated statutory-damages question (treating albums as single “works” for compilations). After Cox v. Sony, the Grande liability rationale appears difficult to sustain: knowledge of specific infringing subscribers, without inducement, is now insufficient. Lower courts confronting pending ISP cases will have to retrofit their analyses to the inducement/non-substantial-use framework.

Open questions

What survives as “inducement” for a neutral conduit? The Court did not map the outer edge of active encouragement. Could aggressive marketing to known piracy communities, or deliberate design choices that facilitate infringement, suffice even for a general broadband provider?
Does § 512 compliance still carry independent weight? If baseline liability is so narrow, what is the practical penalty for a deficient repeat-infringer policy — and will plaintiffs even litigate safe-harbor forfeiture going forward?
How far does the holding reach beyond conduit ISPs? Hosting platforms, cloud providers, and generative-AI services all invoke “substantial noninfringing uses.” The opinion’s logic is technology-agnostic, but its application to storage and platform defendants under § 512(c) is untested.
What happens on remand? With the contributory predicate gone and vicarious liability already rejected below, the realistic path is judgment for Cox — but the formal posture leaves room for further proceedings.

Implications

For ISPs and platforms: Mere awareness of infringing users — even named, repeat offenders — no longer creates contributory exposure. Litigation risk now concentrates on inducement-type conduct and product design.
For rights holders: The “strip the safe harbor, then collect” strategy is substantially weakened. Plaintiffs must now prove intent, not curated notices and knowledge.
For § 512 compliance programs: The repeat-infringer requirement remains the law, but its enforcement value is in question; counsel should still maintain policies while recognizing the downstream liability calculus has shifted.
For pending appeals (including Grande): Verdicts built on knowledge-plus-continued-service are now vulnerable and should be re-examined under the new standard.
For generative-AI defendants: Expect heavy reliance on Cox v. Sony’s reaffirmation of Sony/Grokster dual-use protection.

Frequently asked questions

Did the Supreme Court eliminate the DMCA repeat-infringer policy requirement? No. Section 512(i)‘s requirement that providers reasonably implement a repeat-infringer termination policy remains intact. The Court changed the consequences of losing the safe harbor: forfeiture returns a defendant to ordinary secondary-liability principles, which now require intent rather than mere knowledge.

Was the decision unanimous? The judgment was unanimous — all nine Justices agreed Cox should prevail and the Fourth Circuit be reversed. But the reasoning split 7-2: Justices Sotomayor and Jackson concurred only in the judgment and rejected the majority’s narrowing of contributory liability.

What happened to the $1 billion verdict? It is gone. The Fourth Circuit had already vacated the original award and remanded for a new damages trial; the Supreme Court then reversed the liability finding that supported any damages, because the willfulness finding depended on the now-rejected contributory-infringement theory.