In a published opinion issued on 16 September 2026, the US Court of Appeals for the Ninth Circuit affirmed dismissal of the DMCA copyright-management-information claim in Doe v. GitHub. Based on the complaint’s description of Copilot and Codex, the court treated their output as newly generated work that had never carried the information, rather than copies of existing works from which copyright management information (CMI) had been removed or altered. Similarity to source code and missing attribution therefore did not suffice on the pleaded output theory. The decision is not a general exemption for AI coding tools: the alternative training-input theory was forfeited, two contract claims remain pending, and the court expressed no view on whether the output infringes copyright.
The court also rejected literal identicality as an independent DMCA requirement; substantial reproduction without CMI may still provide strong circumstantial evidence of removal. JCIPO recommends reviewing code provenance, duplicate-code filters and open-source licence scanning, while retaining records of tool settings, outputs and subsequent processing. Rights holders should distinguish CMI removal, copyright infringement and licence breaches, and preserve evidence of the existing copy, the information it carried and how that information was removed or altered. Missing attribution alone does not establish the same legal claim in every case.
