GitHub's Terms of Service Authorized the Use of Others' Code to Train Copilot

ARGUMENT

III. PLAINTIFFS FAIL TO STATE A BREACH-OF-LICENSE CLAIM.

B. To The Extent Count II Is Based On Training Copilot With Code Found In Public Repositories, It Is Foreclosed By GitHub’s Terms Of Service.

In all events, to the extent Plaintiffs’ unexplained breach theory against GitHub is predicated on training Copilot with code published in public GitHub repositories, Count II is foreclosed by GitHub’s Terms of Service. Compl. Ex. 1.

Consistent with open source licensing principles, GitHub’s TOS expressly authorizes the training of Copilot. The TOS informs users that “You own content you create, but you allow us certain rights to it.” Compl. Ex. 1 at 27 (TOS at 6). “These license grants apply to Your Content” when uploaded to GitHub, notwithstanding any other license terms that might be attached. Id. Every GitHub user chooses whether to allow their “repositories to be viewed publicly.” Id. No user is required to make a repository public. If they choose to do so, they “grant [GitHub] … the right to store, archive, parse, and display” publicly posted content and “make incidental copies, as necessary to provide the Service, including improving the Service over time.” Id. This includes the right to “copy” public material “to our database,” “parse it into a search index or otherwise analyze it on our servers,” and “share it with other users.” Id. at 27- 28 (TOS at 6-7). The “Service” includes all “the applications, software, products, and services provided by GitHub,” id. at 24 (TOS at 3), which includes Copilot. See Compl. ¶ 8; Compl. Ex. 1 at 24, 38 (TOS at 3, Copilot Additional Terms).

As Plaintiffs allege, Copilot functions by evaluating the code contained in GitHub’s public repositories, Compl. ¶¶ 82-83, in order to “infer[] statistical patterns governing the structure of code.” Compl. ¶ 52. That activity is squarely encompassed by the contractual authorization, from every public code repository owner, for GitHub to “store, archive, parse, and display” publicly posted content “as necessary to provide … the applications, software, products, and services provided by GitHub” and to “analyze it on our servers.” Compl. Ex. 1 at 27-28, 24 (TOS at 6-7, 3). Even if the Complaint put forward any plausible theory of how inspection and analysis of publicly posted material violated an open source license provision, Plaintiffs’ acceptance of the GitHub TOS forecloses a claim for breach of contract based on Copilot’s training on the public code. Dismissal is proper where a complaint alleges facts that demonstrate the claim is barred as a matter of law. Orchard Supply Hardware LLC v. Home Depot USA, Inc., 967 F. Supp. 2d 1347, 1354 (N.D. Cal. 2013). So it is here.