Source: CCIA

The Ninth Circuit adopted the position urged by CCIA and the Chamber of Progress in Doe v. Github, addressing whether Section 1202(b) of the Digital Millennium Copyright Act can create liability when an AI tool outputs code that resembles copyrighted source code without including the original work’s copyright management information (CMI). The court framed the core inquiry as whether training AI on publicly available code and producing similar code without attribution or embedded CMI falls within Section 1202(b)’s scope.

CCIA and its co‑filers argued that Section 1202(b) prohibits only removal or alteration of CMI from existing copies and does not impose an affirmative obligation to include CMI on newly generated works, warning that a broader reading would chill AI development. The court rejected that expansive interpretation, holding that to state a claim under section 1202 plaintiffs must allege facts showing defendants removed or altered CMI from copies of existing protected works rather than merely alleging a new similar work lacks the original CMI. The opinion quotes that alleging similarity without facts of removal or alteration is insufficient. The ruling is not an across‑the‑board immunity: the court observed that copyright or DMCA liability may still arise where appropriate facts support it, and similarities can still be circumstantial evidence of CMI removal if substantiated. Overall, the decision narrows Section 1202(b)’s application in the AI training context by keeping its focus on anti‑tampering with CMI and avoiding transforming the provision into a broad liability rule for generative AI outputs. The piece also notes CCIA’s role as an international trade association representing communications and technology firms and that CCIA submitted the joint brief to the Ninth Circuit.