Judge narrows GitHub Copilot lawsuit, dismissing most claims

US District Judge Jon Tigar issued a ruling that narrowed a class-action lawsuit against GitHub, Microsoft and OpenAI over Copilot, the AI coding assistant, dismissing most of the plaintiffs’ claims while allowing the case to continue on a reduced basis. The suit, filed by anonymous programmers in November 2022, was one of the first pieces of litigation to test whether training a code-generation model on public repositories infringed the licence terms attached to that code.

Tigar dismissed claims of negligence, intentional and negligent interference with prospective economic relations, and unjust enrichment, largely on the grounds that they were pre-empted by federal copyright law. He also dismissed the plaintiffs’ claim under the Digital Millennium Copyright Act’s provision on copyright management information, finding that the code Copilot had reproduced was not alleged to be sufficiently identical to the originals to support that theory. Two of five anonymous plaintiffs lost standing for damages claims for lack of a sufficiently specific injury.

What survived was narrower but consequential: a breach-of-contract claim resting on the argument that Copilot’s output violated the attribution and licence terms of the open-source code it was trained on, plus damages claims for the three remaining plaintiffs. That claim — whether reproducing licensed open-source code without preserving its licence terms constitutes a breach separate from copyright infringement itself — became the case’s central surviving theory, tested further in a subsequent ruling in mid-2024 that dismissed the DMCA claims outright while letting the licence-breach claim proceed toward a jury.

The order was among the earliest judicial rulings to work through what kinds of legal claims could survive a motion to dismiss in AI training-data litigation, at a point when several similar suits against other developers were still in their opening stages.