TL;DR:
- The Ninth Circuit on September 16, 2026 held that outputs produced by GitHub Copilot and OpenAI Codex are not “copies” from which copyright management information (CMI) was removed or altered, so DMCA Section 1202(b) claims fail. This confirms that AI-generated code outputs are not automatically treated as DMCA violations for removing CMI, even when those outputs omit licensing notes found in training data. The court left intact other theories, including contract claims, but dismissed the DMCA 1202 claim at the pleading stage. This ruling narrows the leverage in DMCA 1202 disputes involving generative AI outputs and shapes the litigation landscape for AI-assisted software development and related discovery battles. Practitioners should reassess DMCA 1202 strategies in AI cases and rely on alternative theories where liability may still exist. The decision is Doe v. GitHub, Inc., Nos. 24-7700, 24-7716, 24-7723 (9th Cir. Sept. 16, 2026). Primary sources and expert commentary are linked below.
What happened in Doe v. GitHub
In Doe v. GitHub, Inc., programmers argued that Copilot and Codex reproduced their open source code without proper copyright management information, violating DMCA Section 1202(b). The district court dismissed the DMCA claims, first with leave to amend and then with prejudice. On appeal, the Ninth Circuit affirmed, holding that the outputs generated by the AI tools are new works, not copies of the original protected code from which CMI was removed or altered. The court explained that “the output theory” does not satisfy § 1202(b)(1) because the AI output does not remove or alter CMI from a copy of an existing work; instead, the tool creates a new work that may or may not infringe copyrights but does not constitute a removal or modification of CMI attached to an existing copy. The court noted that the statute centers on copies that already exist with CMI and requires a removal or alteration of that CMI on those existing copies. This is a key textual distinction, and the court declined to apply an “identicality” requirement beyond the statutory text. These points are detailed in the Ninth Circuit’s opinion and related district court orders. See 24-7700.pdf (Sept. 16, 2026) and accompanying summaries. (cdn.ca9.uscourts.gov)
Practical impact for trial teams
Narrowed DMCA 1202 leverage in AI output disputes. The Doe v. GitHub decision reduces the likelihood that a plaintiff can prevail on a theory that AI-generated outputs constitute “stripped copies” of training data. For trial teams, this means shifting focus away from DMCA 1202 claims against AI-produced outputs toward other theories of liability, including copyright infringement, contract, or licensing disputes, when those issues arise in AI-generated code or content. Practitioners should carefully distinguish between training data concerns and post-training outputs, and structure pleadings accordingly. The ruling also emphasizes the need to gather robust evidence about the exact nature of outputs and licensing terms, rather than relying on general claims about missing attribution alone. See the Ninth Circuit’s formal ruling and practitioner summaries. (cdn.ca9.uscourts.gov)
Influence on defense strategy and pleading choices. Defendants in AI liability cases can cite this ruling to argue that DMCA 1202(b) does not provide a straightforward escape hatch for all AI outputs, particularly when those outputs are newly generated and not straightforward copies of preexisting works. Plaintiffs, in turn, may pivot to contract-based claims, licensing disputes, or direct copyright infringement theories if there is meaningful copying or reproduction of protected material in outputs. The decision signals a narrower path for DMCA-based liability in the AI space, at least in the Ninth Circuit. See also industry commentary from major firms and IP practitioners. (haynesboone.com)
Implications for discovery and evidence handling in AI cases. DMCA arguments often intersect with discovery strategies around training data, model prompts, and outputs. With the Ninth Circuit clarifying that outputs are not “stripped copies” of their training data, litigants may adjust how they request and contest discovery material related to prompts, outputs, and model training. Courts may still require careful handling of sensitive material and licensing terms, but the DMCA 1202 leverage point is narrowed for this factual pattern. For context on the broader AI policy landscape, the federal judiciary has signaled ongoing AI guidance and CM/ECF modernization efforts that will affect how AI evidence and discovery are managed going forward. (uscourts.gov)
How the Ninth Circuit decision fits into the broader AI litigation landscape
The Doe ruling arrives as courts continue to refine the boundaries of liability for AI-driven outputs. While the DMCA framework is limited in this context, other avenues for relief remain actively litigated, including copyright infringement, licensing, and contract claims, as well as emerging standards for AI training data usage. Industry observers and law firms have highlighted the decision as a critical anchor for how courts interpret “removal” or “alteration” of CMI in the AI era, which will guide trial teams in drafting pleadings and planning discovery across similar cases. See practitioner analyses and the published opinion for the core statutory interpretation and the court’s treatment of “input” versus “output” theories. (cdn.ca9.uscourts.gov)
Related developments in AI policy and court practice. In parallel with the Doe decision, the federal judiciary publicly outlined progress on case management modernization and AI policy. The Judiciary’s September 17, 2026 update notes rapid CM/ECF modernization and the ongoing AI task force work, signaling that courts will increasingly encounter AI-enabled practice and evidence in both filings and courtroom sessions. This broader context matters for trial teams planning for AI-related discovery, evidence authentication, and litigation strategy in the near term. (uscourts.gov)
Practical steps for litigators in the wake of the decision
Reassess DMCA 1202 strategies in AI cases. When facing claims involving AI-generated outputs, scrutinize whether CMI was actually removed or altered from an existing copy, rather than assuming outputs are impermissible under DMCA 1202(b). If the case turns on training data usage, consider preserved claims or alternative theories rather than relying solely on §1202(b). (cdn.ca9.uscourts.gov)
Build a multi-theory pleadings plan. The Doe decision underscores the importance of tying claims to concrete actions and existing copies. Lawyers should prepare parallel paths for potential liability, including copyright infringement, licensing violations, breach of contract, or misuse of trade secrets where applicable. Strengthen factual records around how AI systems were trained, what outputs were produced, and what licensing terms applied to those outputs. (cdn.ca9.uscourts.gov)
Prepare for AI-enabled discovery realities in courtrooms. With the judiciary actively pursuing CM/ECF modernization and AI policy, trial teams should anticipate evolving evidence handling, prompt discovery protocols, and potential privacy safeguards around model prompts and outputs. Proactive planning now can reduce friction in later stages of a case and improve the ability to demonstrate trial-readiness and credibility in front of judges. (uscourts.gov)
Leverage Objection Academy for AI-rights readiness. In evergreen practice contexts, Arm practitioners with robust objection drills, evidence training, and courtroom simulations tailored to AI and digital-evidence issues. Objection Academy can help trial teams rehearse cross-examinations and objections related to AI-generated outputs, licensing terms, and training-data disclosures, contributing to faster, more precise courtroom performance in tech-forward cases.
Conclusion
The Ninth Circuit’s Doe v. GitHub decision marks a decisive clarification on how DMCA Section 1202 applies to AI-generated outputs. By treating outputs as new works rather than copies from which CMI was removed or altered, the court narrows a common DMCA-based litigation strategy in AI matters. For trial attorneys, the development calls for disciplined pleading, careful theory selection, and proactive discovery planning while staying aligned with broader federal AI-guidance initiatives. As courts and technology continue to intersect, tools and training that enhance courtroom readiness, such as Objection Academy, become increasingly valuable in translating evolving legal standards into practical, trial-ready skills.
Sources:
- 24-7700.pdf, Doe v. GitHub, Inc., Ninth Circuit Court of Appeals (Sept. 16, 2026). Primary court opinion detailing the “output theory” and the statutory interpretation of DMCA §1202(b). (cdn.ca9.uscourts.gov)
- Doe v. Github, Inc., 24-7700 (9th Cir. Sept. 16, 2026) Justia summary. (law.justia.com)
- Judiciary Cites Progress on Case Management, Property Authority, and AI, U.S. Courts News Release (Sept. 17, 2026). Broad context on AI policy and CM/ECF modernization affecting litigation workflows. (uscourts.gov)
- Haynes Boone summary of the Ninth Circuit ruling on DMCA §1202 in Doe v. GitHub (Sept. 2026). (haynesboone.com)