Generative artificial intelligence programs now are capable of generating software code based on prompts entered by users. As with all technology powered by generative neural networks, these programs learn how to perform their tasks by using training data. Copilot, a generative AI add-on for programming platforms, was trained using source code that may be subject to open-source licenses. Doe v. Github Inc. et al. is a currently ongoing lawsuit that will address the copyright issues inherent in software technology like Copilot and other generative AI programs.
Panitch Schwarze attorney Jeffrey W. Gluck, Ph.D. authored an article in the Journal on Emerging Issues in Litigation examining the details of the case. He notes that Github, the company that developed Copilot, runs an open-source depository containing software that is subject to permission to be copied and reused, usually under the conditions of an open-source license. The plaintiffs in this case assert that Copilot was trained using their source code and that Github did not adhere to the various open-source licenses they used, which required, at the very least, attribution of the code to the developer, as well as retaining the licensing terms with the code.
Gluck explains that the complaint alleges violations of the Digital Millennium Copyright Act, which specifically prohibits intentionally removing or altering any copyright management information. Although no direct violations of copyright were asserted in the complaint, other pending lawsuits have made copyright infringement claims in connection with generative AI. With the recent unprecedented progress of generative AI technology, the landscape of intellectual property law is sure to evolve in the coming years.
Read the full article here: Copyright Issues in Generative AI for Software: Doe v. Github, Inc. et al.