TL;DR
The Ninth Circuit ruled that AI coding tools do not automatically violate the DMCA by generating code without copyright attribution, handing GitHub, Microsoft, and OpenAI a meaningful but narrow legal win.
What happened
- The Ninth Circuit issued its ruling on September 16, 2026 in Doe v. GitHub, the first major appellate decision on AI coding tools and open-source code.
- Anonymous open-source developers sued GitHub, Microsoft, and OpenAI, arguing that GitHub Copilot and OpenAI Codex generated code stripped of author attribution, copyright notices, and license terms from public GitHub repositories.
- The court ruled that generating a new work without copyright management information (CMI) does not equal unlawfully removing CMI under Section 1202 of the DMCA.
- Judge Eric Miller wrote that "one who creates a new work and fails to include CMI cannot be said to have 'removed' or 'altered' anything."
- Most other claims were dismissed on procedural grounds: the complaint, the court noted, "is not about training. It just isn't."
Why it matters
- Reduced DMCA litigation risk for every company shipping AI code assistants, general-purpose LLMs, and agentic development tools, not just Copilot.
- The ruling does NOT resolve whether training on open-source code is copyright infringement, whether such training qualifies as fair use, or whether output that closely reproduces protected code infringes copyright.
- Open-source license obligations remain live: the district court has yet to decide whether developers' MIT, GPL, Apache, and other licenses were honored, a claim the OSI calls "unambiguous."
- The Software Freedom Conservancy warns the decision is easy to overread: it turns on how narrowly the plaintiffs framed their claims, not on a broad endorsement of AI training practices.
- A better-argued future case could reach a different outcome, especially if a system behaves more like a search engine returning identical code with CMI stripped.
What to watch next
- District court ruling on open-source license compliance: whether Copilot and Codex honored MIT, GPL, and Apache attribution requirements is still undecided and could produce a far more consequential outcome.
- A new, more precisely pleaded lawsuit targeting the training data ingestion phase directly, which this court explicitly left unaddressed.
- Legislative or regulatory movement on AI training and open-source rights, as courts continue to signal that Congress, not judges, may need to draw these lines.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.