Ninth Circuit rejects a DMCA claim against GitHub Copilot

The US Court of Appeals for the Ninth Circuit ruled in Doe v. GitHub, Inc. that AI-generated code lacking copyright-management information does not, by itself, establish unlawful removal of that information under section 1202(b) of the Digital Millennium Copyright Act. The decision affirmed part of a district-court ruling that had already dismissed most of the same lawsuit against GitHub, Microsoft and OpenAI over Copilot and Codex.

The plaintiffs, anonymous programmers who had published code on public GitHub repositories, argued that Copilot sometimes reproduced “essentially verbatim” copies of their code without the attribution and licence notices that accompanied the originals — an “identicality” theory under which the mere absence of that information from the output was itself evidence it had been stripped out. Judge Eric Miller, writing for the panel, rejected that reading of the statute: section 1202(b) targets removing or altering copyright-management information from an existing copy of a protected work, not generating a new work that never carried the information to begin with. The court did find the plaintiffs had Article III standing to bring the claim at all, and declined to reach a separate “input” theory of DMCA liability that it held had been forfeited on appeal.

The ruling is, according to the Electronic Frontier Foundation, which filed an amicus brief in the case, the first circuit-level precedent addressing this DMCA theory as applied to AI-generated code, and narrows one of the legal avenues plaintiffs have used against AI code-generation tools. It leaves intact the case’s separate breach-of-contract claim — that reproducing licensed open-source code without preserving its licence terms is itself a breach independent of copyright law — which continues toward trial in the district court.