In the lawsuit J. Doe 1 et al. vs GitHub, the plaintiffs claim that Microsoft, OpenAI, and GitHub, through their commercial AI-based systems, OpenAI's Codex and GitHub's Copilot, stole their open-source code. According to the class action suit, AI-generated code often consists of nearly identical strings of code collected from public GitHub repositories, without proper open-source license attributions. The rise of generative AI presents great potential in the field of engineering but also poses challenges for enterprises and engineers as they navigate the impact of AI on their roles, strategies, data, solutions, and product development. ZDNET explores the future roadmap for integrating generative AI into software development from various perspectives. Read more. However, this issue extends beyond Microsoft alone. Sean O'Brien, a cybersecurity lecturer at Yale Law School and founder of the Yale Privacy Lab, believes that there will soon be a new form of trolling similar to patent trolls, focused on AI-generated works. As more authors use AI-powered tools to release code under proprietary licenses, a feedback loop is created, leading to software ecosystems being contaminated with proprietary code that may face cease-and-desist claims from opportunistic firms. On the other hand, some, like German researcher and politician Felix Reda, contend that all AI-produced code should be considered public domain. The question then arises: What steps should be taken?Simply claiming that an AI is open source is not sufficient. For example, Meta claims that Llama 2 is open source, but it is not. Erica Brescia, a managing director at RedPoint, a venture capital firm supportive of open source, raises concerns on Twitter about the justification of calling Llama 2 open source when it does not adhere to an OSI-approved license or comply with the OSD. It appears that Meta is using "open source" more as a marketing term than a legal one. This approach will likely prove ineffective as lawsuits accumulate. The specific problem with Llama 2 is that it prevents highly profitable companies from using it.

According to Stephen O'Grady, an expert on open-source licensing and co-founder of RedMonk: imagine if Linux were only open source to those who worked at Facebook. Additionally, Amanda Brock, CEO of OpenUK, suggests that significant AI or LLM (large language models) may no longer be licensed as open source because the key element of open source is the Open Source Definition. The road to this definition has been long and challenging. The origins of free software licenses date back to the early 1980s when MIT Lab programmer Richard M. Stallman was unable to access and modify the source code of an early laser printer, the Xerox 9700. Stallman created the GNU General Public License (GPL), which, along with the Berkeley Software Distribution (BSD) license, played a significant role in the formation of free software licenses. The phrase "open source" emerged as a substitute for "free software" when Mozilla's source code was released by Netscape, leading to the coining of the term by influential figures such as Eric S. Raymond, Bruce Perens, and others. In 1998, Perens and Raymond founded the OSI, which developed the Open Source Definition (OSD) as a guide for all open-source licenses. Compliance with the OSD is mandatory for any open-source license. However, applying open-source licenses to AI and LLMs poses greater difficulty. The implications of combining software and data in AI/ML artifacts, such as datasets and models, create challenges for existing open-source licenses. Stefano Maffulli, executive director of the Open Source Initiative (OSI), pointed out that this issue emerged with the release of GitHub Copilot two years ago, which fueled the considerable harvesting of code for machine learning. To address this, a new definition for open-source AI is needed, one that all stakeholders can agree upon and work with. Maffulli, along with industry leaders from Google, Microsoft, GitHub, Open Forum Europe, and many others, is collaborating on a draft to establish a common understanding of open-source AI. Although this initial draft will be released soon, it should be finalized promptly due to the fast-paced advancement of AI. The involvement of various AI players in this defining process underscores the importance of establishing an open-source framework for AI.