Lawsuit tracker

Doe v. GitHub (Copilot)

Doe 1 et al. v. GitHub, Inc., Microsoft Corp. and OpenAI entities, No. 4:22-cv-06823 (N.D. Cal.); interlocutory appeal pending (9th Cir.)

PlaintiffsAnonymous open-source developers (proposed class)
DefendantsGitHub, Inc.; Microsoft Corporation; OpenAI, Inc. and related OpenAI entities
CourtU.S. District Court, Northern District of California (Judge Jon S. Tigar); interlocutory appeal before the U.S. Court of Appeals for the Ninth Circuit
Filed2022-11-03
StatusOn appeal
Content typecode
Last updated2026-07-21
Verified against the cited sources at last update. Litigation moves fast — check the sources below for the current status.

The claims

DMCA §1202(b) removal of copyright management information; breach of open-source license terms (contract); related state-law claims, most now dismissed

What has happened

The first U.S. class action over generative AI training. Anonymous developers sued GitHub, Microsoft and OpenAI in November 2022, alleging Copilot and the Codex model were trained on public repositories and emit code stripped of the license text, attribution and copyright notices that open-source licenses require. Notably, the plaintiffs pleaded no direct copyright infringement count; the case runs on DMCA §1202(b) and contract theories. Judge Tigar trimmed the case in successive orders. In June 2024 he dismissed the §1202(b) claims with prejudice, reading the statute to cover only identical copies of a work, while letting contract claims over open-source license terms proceed. In September 2024 he certified the identicality question for interlocutory appeal, and the Ninth Circuit accepted it. Amici split sharply: IP professors backed GitHub’s reading, while the News/Media Alliance urged the court to reject an identicality requirement. The panel heard argument on February 11, 2026.

Key developments

  • 2022-11-03Proposed class action filed in the Northern District of California over Copilot and Codex training on public GitHub code.
  • 2023-05First dismissal order trims most claims; plaintiffs replead.
  • 2024-06Court dismisses the DMCA §1202(b) claims with prejudice, holding the statute reaches only identical copies; open-source license contract claims survive.
  • 2024-09Judge Tigar certifies the identicality question for interlocutory appeal; the Ninth Circuit takes it up.
  • 2026-02-11Ninth Circuit hears argument on the §1202(b) identicality question. Decision pending.

Why it matters for training data

§1202 is the main federal hook plaintiffs have against training pipelines that strip licenses and attribution, because it does not require proving the output infringes. If the Ninth Circuit affirms the identicality requirement, CMI claims over training data mostly die in that circuit, and code-related exposure shifts to contract: open-source license compliance, attribution and notice terms. If it reverses, any dataset that strips headers, licenses or bylines carries per-work statutory-damages risk. Either way, suppliers should preserve provenance and license metadata, and buyers should ask whether it was preserved.

Sources

Deeper analysis

← All case files Weekly lawsuit news

Want data that clears this in diligence?

Whether you're building a model or sitting on an archive, the first conversation is short and specific.

Send a brief