UF law scholars: Supreme Court’s Cox ruling quietly shields AI from copyright liability

The Supreme Court’s decision in Cox Communications v. Sony Music Entertainment is a welcome guardrail against a legal theory that could have strangled AI development in its infancy. The ruling reaffirms a foundational principle of internet law: platforms and providers should not be held financially responsible for every act of infringement committed by their users. That principle, first codified in the Digital Millennium Copyright Act’s safe-harbor provisions, made the modern internet possible — and the same logic must extend to AI systems if the technology is to flourish.

AI companies do not set out to infringe anyone’s copyright. Like a cable provider that cannot monitor every byte traveling its network, an AI developer cannot realistically police every output generated by millions of users in real time. Holding AI firms to a standard of strict vicarious liability — essentially making them insurers against any downstream misuse — would impose compliance costs so staggering that only the largest incumbents could survive them. The likely result would be a consolidation of AI power into a handful of well-capitalized corporations, the opposite of the open and competitive ecosystem that drives genuine innovation.

UF law professors Bambauer and Nguyen are right to recognize the decision’s significance beyond its music-industry context. Copyright law was designed to incentivize creation, not to be weaponized as a market-entry barrier. When the recording industry pursued billion-dollar damages against an internet service provider for its subscribers’ file-sharing, it was pursuing a legal theory that would let rights-holders extract rent from any technology they disliked, regardless of that technology’s own legitimate uses. The Court’s rejection of that theory preserves the breathing room that technologists need to experiment, iterate, and build.

The practical stakes are enormous. AI-assisted medicine, legal research, scientific discovery, and creative tools all depend on the ability to train models on large datasets and deploy them broadly. A legal regime that treated every ambiguous output as a potential nine-figure liability would chill every one of those applications. Protecting the infrastructure of AI — as the Cox ruling effectively does — is not a gift to corporations; it is a precondition for the public benefits the technology promises.

It is telling that the creative industries reacted to the Cox Communications ruling with alarm, because what the Supreme Court has done — however inadvertently — is weaken one of the few legal mechanisms available to artists, musicians, and writers who are watching their livelihoods eroded by AI systems trained on their work without consent or compensation. Framing this outcome as a triumph for innovation papers over a serious injustice.

Copyright exists precisely to give creators control over how their work is used and distributed. When an AI company trains a model on millions of songs, novels, or images scraped from the internet, it is deriving direct commercial value from those works. The Cox ruling, as Bambauer and Nguyen acknowledge, reduces the liability exposure AI companies face when their users generate infringing content — but the more fundamental problem is that the training process itself may constitute large-scale infringement. A ruling that insulates downstream distribution does nothing to address the upstream extraction, and celebrating it as a shield for AI ignores the harm inflicted at the point of ingestion.

Moreover, the analogy between a cable company passively carrying data and an AI company actively using creative works to build a commercial product is strained. Cox did not transform the music it carried; it transmitted it. AI companies transform, recombine, and reproduce copyrighted expression in ways that compete directly with the original creators. A songwriter whose style is replicated by an AI trained on her catalog has not merely had her work transmitted — she has had it absorbed and commodified. The legal framework that governs passive conduits should not automatically govern active learning systems.

The music industry’s anger is not protectionism or technophobia. It is a reasonable response to a legal environment that has consistently lagged behind technological reality, leaving creators exposed while platform companies profit. If AI is to be a genuine public good rather than a mechanism for transferring value from human creators to corporate shareholders, courts and Congress will need to craft rules that go beyond the passive-carrier logic of the Cox decision — rules that require AI developers to seek licenses, share revenue, or at minimum provide meaningful opt-out rights. A ruling celebrated for what it does not say is a thin foundation for a just copyright regime.