TL;DR: The Ninth Circuit in Doe v. GitHub, Inc., No. 24-7700 (Sept. 16, 2026) held that DMCA Section 1202 claims fail for the outputs of generative AI tools. The court concluded that Copilot and Codex generate new works rather than copies of existing code, so they do not “remove or alter” copyright management information (CMI) from copies. The district court’s dismissal of the DMCA claims was affirmed, and the court declined to consider the plaintiffs’ forfeited input theory. Practically, this ruling reduces the DMCA 1202 risk for AI-generated outputs in litigation, while leaving open questions about training-data theory and future developments in other jurisdictions. (cdn.ca9.uscourts.gov)
What happened
Doe v. GitHub, Inc. involves programmers who published open-source code and alleged that GitHub Copilot and Codex produced verbatim or near-verbatim output without attributing CMI, allegedly violating DMCA § 1202(b). The Ninth Circuit affirmed the district court’s dismissal of the DMCA claims, but on distinct grounds. The court held that the plaintiffs could plausibly show standing, but the DMCA claim under the “output theory” fails because the AI’s output is not a copy of an existing work with CMI that was removed or altered; instead, the AI creates new works that did not contain the CMI in the first place. The court also declined to consider an alleged “input theory” because the plaintiffs forfeited that theory on appeal. In short, the controlling rule is that generation of new works by AI that do not carry the original CMI does not violate § 1202(b). (cdn.ca9.uscourts.gov)
Key passages from the decision explain the distinction between the theories and the statutory text. The court began with the statutory framework: DMCA § 1202(b) prohibits removing or altering CMI, and prohibits distributing works or copies when CMI has been removed or altered. The panel then analyzed the two theories: the input theory (removal of CMI during training) and the output theory (AI output lacking CMI). It held the input theory forfeited and rejected the output theory on the merits because the output is not a copy of the underlying work bearing CMI; it is a newly created work that does not contain the CMI to begin with. The Ninth Circuit thus affirmed dismissal of the DMCA claims under 1202(b). (cdn.ca9.uscourts.gov)
For practitioners, the decision clarifies that at least as to DMCA § 1202(b) claims, the mere production of AI-generated outputs without CMI does not automatically equate to removing CMI from a copy of a work. This has immediate implications for cases in which parties contend that AI tools reproduce protected material with missing attribution. The Ninth Circuit’s analysis centers on the text of § 1202(b) and the nature of the AI output, not on broader arguments about AI training or model behavior. (law.justia.com)
Practical implications for trial teams
Risk posture for DMCA 1202 in AI-output disputes. The decision signals that courts may treat AI-generated outputs as new works, not as copies that have had CMI removed. This reduces the likelihood of a § 1202(b) liability bite based solely on the absence of attribution in AI-produced outputs. Attorneys litigating IP or licensing disputes involving AI tools should assess whether a § 1202(b) claim is still viable under an input theory or other theories, but should recognize that the “output” approach may not survive as a DMCA claim in this jurisdiction. (cdn.ca9.uscourts.gov)
Training-data claims remain unsettled. Although the court declined to reach the input theory on the merits due to forfeiture, the district court’s and the Ninth Circuit’s emphasis on procedural posture means that training-data attributions and licensing concerns can still be live issues in other cases or in other circuits. Litigants pursuing training-data or attribution theories should monitor developments in other jurisdictions and consider tailoring motions and pleadings to preserve or preserve-like theories where permissible. (cdn.ca9.uscourts.gov)
Standing and injury-in-fact in AI cases. The Ninth Circuit acknowledged Article III standing for a plausible risk of injury from AI outputs, even while disposing of the DMCA claim on the merits. This combination means that plaintiffs may still proceed with related contract, licensing, or privacy/consumer-protection theories, while DMCA-based claims may be narrower in the Ninth Circuit for AI-generated outputs. Practitioners should structure pleadings to address standing carefully when AI players and training data are involved. (law.justia.com)
Discovery strategy and evidence handling. With the DMCA 1202 landscape clarified in this ruling, litigants focusing on AI-generated evidence should adjust discovery plans accordingly. Requests narrowly targeting CMI removal from existing copies may be less likely to succeed in courts applying the Doe v. GitHub framework for outputs. Conversely, parties should consider robust preservation and authentication strategies for AI outputs, including chain-of-custody for prompts and system configurations if they intend to rely on such outputs at trial. Practitioners should work with evidence teams to ensure that the evidentiary record faithfully reflects how the AI-produced material was generated and used. (cdn.ca9.uscourts.gov)
Objection handling and trial-readiness. In a world where AI outputs increasingly appear in evidence, objection handling becomes essential. Training in how to challenge the admissibility, authenticity, or reliability of AI-generated outputs is increasingly core to trial practice. Objection Academy provides practical drills and simulations focused on objection commands, foundation, authentication, and real-time courtroom objections, supporting trial teams facing AI-driven exhibits and prompts. Incorporating targeted practice on AI-generated content can help attorneys anticipate issues before trial and streamline effective advocacy. This aligns with a broader emphasis on developing trial-readiness skills in a fast-evolving evidentiary landscape. Practitioners may want to pilot objection-focused training as part of their pretrial preparation.
Practical next steps for practitioners
Reevaluate DMCA 1202 strategies in open IP/AI cases within the Ninth Circuit footprint. If facing AI-generated outputs, assess whether any § 1202 claims hinge on an input theory or on altered copies, and tailor pleadings and motions accordingly. Stay alert for similar rulings in other circuits, as the doctrinal splits on DMCA 1202 and AI remain fluid. (cdn.ca9.uscourts.gov)
Strengthen non-DMCA theories. Since the DMCA angle may be narrowed in some jurisdictions, fortify contract, licensing, or misappropriation theories related to training data, attribution, or use of open-source licenses. Courts may view these as independent or alternative paths for relief, especially in cases involving open-source code and AI-assisted development. (law.justia.com)
Prepare for cross-jurisdictional strategy. Although the Ninth Circuit’s ruling is influential in its circuit, open cases in other jurisdictions may reach different conclusions. Counsel should coordinate with national teams to harmonize strategy and preserve options across forums. (law.justia.com)
Invest in trial-readiness training. Even when DMCA claims are limited, the increasing prevalence of AI-generated material in litigation makes objections, authentication, and evidentiary handling of AI outputs a practical competency for trial teams. Objection Academy’s resources on objections to AI-driven evidence and courtroom simulations can be a valuable component of a comprehensive pretrial plan. Practicing these skills can reduce trial risk and improve advocacy when AI outputs surface during discovery or at trial.
Significance for the current year
Doe v. GitHub marks a pivotal point in the legal treatment of AI-generated outputs under the DMCA. The Ninth Circuit’s focus on the distinction between copies with removed CMI and newly created outputs provides a clear framework for analyzing DMCA § 1202 claims in AI contexts. While the case narrows the viability of certain DMCA theories in this circuit, it also highlights the ongoing importance of preserving robust, independent claims in IP and licensing disputes involving AI. For trial teams, the decision reinforces the need to align evidentiary strategy with the narrow contours of DMCA liability while actively preparing to handle AI-generated material in the courtroom through rigorous objections, authentication, and trial-readiness training.
Sources:
- Doe v. GitHub, Inc., No. 24-7700 (9th Cir. Sept. 16, 2026), PDF of the opinion. (cdn.ca9.uscourts.gov)
- DOE v. GITHUB, INC., 24-7700, Justia Opinion Summary (Sept. 16, 2026). (law.justia.com)
- The Ninth Circuit’s discussion and analysis of the output theory and forfeiture of the input theory. (cdn.ca9.uscourts.gov)
- Additional analysis and commentary on the ruling and its implications for AI, DMCA, and litigation practice. (ropesgray.com)
End of article