GitHub’s Copilot has been sued for widespread software piracy in a major lawsuit. The AI tool allegedly violates the licensing rights of developers who publish their software on GitHub under open-source licenses.
According to the law firm representing the software developers, Copilot abuses the rights of the software used for its functionality. The OpenAI-based tool provides developers with code suggestions based on their project needs. Users pay between 10 and 100 euros a year. OpenAI was also sued in the case.
The plaintiffs filed their complaint in a federal court in California, alleging that Microsoft’s GitHub violated terms and conditions, privacy policies, the California Consumer Privacy Act and DMCA Code 1202, which prohibits the removal of copyright information.
GitHub removes licenses
GitHub Copilot allegedly trains its AI models through public repositories available in GitHub. This involves code that other developers have published on GitHub under open-source conditions.
These include eleven open-source conditions that require users to credit a developer’s name and copyright, such as the MIT license, GNU General Public License and Apache license.
In the case of GitHub Copilot, the licenses of thousands of developers are simply removed, the complaint says. The tool allegedly commits software piracy at scale.
GitHub Copilot runs entirely on Azure and reproduces code from open-source libraries and repositories. The code used never acknowledges the original developers. Therefore, the tool violates the licenses associated with the original code.
The plaintiffs argue that the service makes open-source code available behind a paywall. Licenses are thereby violated. Open-source code is monetized for personal gain, despite GitHub’s promise never to do so.
In response, GitHub pointed at new functionality announced for Copilot on November 1. When the tool presents a code snippet, it should also offer developers an inventory of equivalent code that can be found in GitHub repositories. Developers should be able to organize this inventory using filters for contribution date, repository license and more.
In addition, GitHub says it’s been transparent about the fact that the tool can produce copied code in some cases. Moreover, the current version reportedly prevents suggestions that match existing code in public repositories. GitHub further detailed how it handles developer copyrights in a blog post.