What it does The opinion addresses an interlocutory appeal under 28 U.S.C. § 1292(b). It holds that the programmers plausibly alleged Article III standing because their complaint described a substantial risk that Copilot would reproduce their code without attribution. On the merits, however, the court concluded that Copilot and Codex generate new works through a probabilistic process rather than retrieve stored copies and strip their copyright management information. The court declined to consider the plaintiffs’ separate “input” theory because it was forfeited. It explained that “identicality” is not an independent statutory element, although substantial reproduction with missing attribution may support an inference of removal.

Who it affects The decision directly affects the pseudonymous programmers and GitHub, Microsoft, and OpenAI, and provides guidance for DMCA claims involving generative AI coding tools and open-source licenses. It does not resolve the plaintiffs’ two contract claims, which remained pending in the district court, or decide ordinary copyright infringement.

Why it matters The ruling limits the use of DMCA § 1202(b) to transform allegations that an AI system generates similar or derivative code without attribution into claims for removal or alteration of copyright management information. It distinguishes those allegations from copying an existing work whose attached attribution information was affirmatively removed.

Key dates and numbers

• Filed September 16, 2026; argued February 11, 2026.
• Ninth Circuit No. 24-7700; ND Cal. district court No. 4:22-cv-06823-JST.
• Relevant provisions: 17 U.S.C. § 1202(b), § 1202(c), and 28 U.S.C. § 1292(b).
• GitHub’s duplicate-detection feature identifies verbatim public-code snippets of 150 characters or more.