The complexities surrounding digital copyright and contract law in the context of open-source software and artificial intelligence were proven in a recent decision by the United States District Court for the Northern District of California, significant developments have unfolded in the case involving Plaintiffs J. Doe 1, et al., against Defendants GitHub, Inc., Microsoft Corporation, and OpenAI, Inc.
The plaintiffs, a group of software developers, brought forward multiple claims against the defendants, centering on allegations of digital copyright infringement and breach of contract related to open-source licenses. The plaintiffs’ second amended complaint (SAC) presented three primary claims: a violation of DMCA Section 1202(b)(1) and 1202(b)(3) against all defendants, breach of contract for violation of open-source software (OSS) licenses against all defendants, and breach of contract for selling licensed materials against GitHub exclusively.
The litigants’ DMCA claims lie in the alleged failure of the defendants to comply with the “identicality” requirement of Section 1202(b). This section prohibits the removal or alteration of copyright management information (CMI) without authorization. The court found that the plaintiffs did not sufficiently demonstrate that the works in question were identical to those protected under the DMCA, a crucial element for the claim to proceed. Consequently, the court dismissed the DMCA claims with prejudice, indicating that the plaintiffs would not be allowed to amend this claim further.
In their SAC, the plaintiffs introduced new assertions to bolster their remaining claims. They pointed to the introduction of a user-settable Copilot filter by GitHub, designed to block or allow code completion suggestions that match publicly available code. The plaintiffs argued that this feature, being optional for users but not for licensors, potentially violates open-source licenses and increases the risk of their code being used without proper attribution or adherence to licensing terms.
Moreover, the plaintiffs referenced recent academic research suggesting that the likelihood of verbatim emission of their code by AI models is increasing. This assertion is based on studies indicating that as AI models like those developed by OpenAI grow in capacity and complexity, the chances of memorizing and reproducing training data verbatim become higher. This risk forms the basis of the plaintiffs’ argument for a breach of contract, claiming that the defendants’ use of their code in AI training without proper licensing adherence constitutes a contract violation.
The court’s decision on the motions to dismiss was mixed. While it dismissed the DMCA claims, it allowed breach of contract claims related to open-source licenses to proceed. This ruling underscore the importance of adhering to the specific terms of open-source licenses, where some of them (the copyleft ones) include requirements for attribution and restrictions on commercialization. The court recognized the potential for significant economic and legal impacts when such licenses are allegedly violated by deciding to allow these claims to proceed.
In addition, the court dismissed the petitioners’ claims for punitive damages and unjust enrichment. The court noted that punitive damages are generally not recoverable in breach of contract cases unless the breach also constitutes tort, which was not sufficiently alleged in this case. Furthermore, the plaintiffs’ failure to address the defendants’ arguments for dismissing the punitive damages claim in their opposition was also a factor in the court’s decision.
The court’s rulings in this case may set important precedents for how similar cases are approached in the future, particularly regarding the interpretation and enforcement of open-source licenses.
For the full decision, click here.
Are you leveraging AI-assisted coding tools in your development process? Don’t let legal uncertainties hold back your innovation. Check out fossity.com for a confidential, affordable, and prompt open-source audit today.