Introduction

The widespread integration of Generative AI technologies (GenAI) has undoubtably shifted the understood workflow in various industries, some more than others. One industry already seeing significant change is the coding industry which, in the last two years has seen the widespread adoption of Microsoft’s GitHub CoPilot as a coding assistant. According to Microsoft, in the two years since its release, over 77,000 companies have adopted CoPilot into their day-to-day operations, allowing CoPilot to guide a significant amount of “mundane boilerplate coding” and debugging (Bousquette, 2025). As companies progressively allow GenAI to undertake more and more of its coding, this begs some very serious questions regarding the implications of AI-generated code on intellectual property rights, product copyright eligibility, and innovation. This essay will first examine the potential risks of CoPilot and similar GenAI tools in terms of proprietary software and ownership disputes, before exploring the broader policy implications, including the urgent need for clearer legal guidelines on GenAI-assisted authorship and copyright eligibility. Finally, it will close with a discussion on possible effects on industry coders– questioning whether reliance on AI will lead to a more homogenous coding landscape that could stifle the creative problem-solving that has long been a hallmark of the industry.

Legal Grey Areas in GenAI-Assisted Coding

CoPilot has seen widespread adoption because of its ability to increase productivity and free up developer time for more complex problem solving. However, this increased integration of GenAI tools raises concerns about intellectual property risks — particularly if AI models were ever to retain and learn from the proprietary code they process. While CoPilot itself only accepts programming related queries, it is powered by GPT-4o, so I posed this question to ChatGPT-4o instead, asking if it, as a whole, learns from individual user interactions. ChatGPT (4o, 2025) responded, “AI models like me do not learn from individual chats in a way that updates my general knowledge for other users […] However, within our ongoing conversations, I can remember details specific to you if they’re relevant to your preferences, projects, or interests.” So while this technology might not currently exist or be in use, it is not improbable to speculate a future where such a process could be quietly rolled out, as the companies behind these GenAI models have been reporting as far back as July 2024 that they are running out of quality data to train models (Tremayne-Pengelly, 2024). If that were to happen, companies could face serious legal and ethical dilemmas: Could proprietary code unknowingly influence another company’s GenAI assisted development, who will be held liable for copyright or IP infringement, and does either company own their code if it was generated by CoPilot or other GenAI programs?

That final dilemma opens the door for a much broader issue regarding US copyright law in the age of GenAI. Currently in the US, as decided by Thaler v. Perlmutter (2023), works generated by AI are exempt from copyright eligibly as US copyright law has repeatedly been emphasized to require human authorship. While the US Copyright Office (2023) has stated that a work with sufficient human authorship is eligible for copyright, it also clarified that “[i]n these cases, copyright will only protect the human-authored aspects of the work, which are ‘independent of’ and do ‘not affect’ the copyright status of the AI-generated material itself.” This clarification seems to generate its own slew of questions regarding what level of human authorship is required (what does sufficient entail?), what level of GenAI involvement disqualifies something from being copyrightable, and how do company’s proprietary code and software fit into this?

US law has already been adapted to allow companies to file their own copyrights, legally classifying them as “people” to fulfill the human authorship requirements. While it is possible–and likely–that GenAI will eventually be classified as a tool, similar to Photoshop or other editing softwares that companies can use to create intellectual property, the fact that GenAI’s classification is currently up in the air puts a significant amount of companies in potential “hot water” should its use be deemed inappropriate by US copyright law. While I do think that outcome is not necessarily likely, the fact remains that ownership over GenAI output remains a highly contentious issue that does not have a conclusive answer yet, as I discussed last week (Hagen, 2025). Ongoing court cases challenge GenAI’s inherent legality, with creators arguing that its very existence violates their intellectual property rights and should disqualify any GenAI-assisted work from copyright protection.

However, even if US Copyright Law was edited to clarify the amount of AI use allowed for copyright eligibility, or even if they decided that AI in general does not disqualify products from copyright eligibly, that still does not provide a solution to the initial dilemma posed by this essay regarding potential unintentional IP infringement by GenAI and who bares the responsibility in such a case. This issue remains unresolved, and I don’t believe a definitive answer will emerge until the ongoing court cases challenging GenAI content ownership reach their verdicts. Regardless, this adds yet another layer of complexity to the use of GenAI in intellectual property creation that creators and companies need to be aware of when considering the use of GenAI. The current legal landscape for US copyright eligibility remains so uncertain regarding GenAI use, that while I would argue that companies likely hold the legal advantage–given the widespread adoption of GenAI even following the Thaler v. Perlmutter (2023) verdict–it is still a debate that companies need to take into account when integrating GenAI into their coding frameworks as the current legal uncertainties could evolve into serious risk.

GenAI influence on Technological Innovation

Having discussed the potential legal and policy implications for widespread GenAI adoption, a final question remains: should coders and companies even want this? Coding has always been a resourceful and innovative field, where creative problem-solving drives progress. Take for example, the Pandas library in Python. This library was created due to a need for more sophisticated data analysis tools in Python, and while solutions did exist in libraries like NumPy, they were cumbersome and inefficient (Pandas, 2025). Pandas emerged out of necessity, filling a gap that existing tools could not. But if GenAI had existed in 2008 and developers could rely on GenAI to quickly generate quick but inefficient NumPy-based solutions, would Pandas have ever been created? This highlights my primary concern: by allowing GenAI to handle tedious or “mundane boilerplate” coding tasks, do we risk fostering complacency in the industry and stifling the creation of innovative solutions?

Yes, GenAI can generate code faster than a human coder, but it lacks creativity. It is simply an amalgamation of previously known code and (for now) does not possess the drive to invent new approaches to old problems. I won’t pretend that I don’t use GenAI tools for debugging, but a small part of me worries that as more coding is offloaded to AI, we risk moving toward a homogenous industry where everything is done the same way. One of the strengths of coding is that there are multiple ways to solve each problem, and it is through these differences in style and approach that new breakthroughs emerge. It would be a shame to lose that diversity to GenAI homogeny.

Conclusion

Overall, while Microsoft’s GitHub CoPilot has seen widespread adoption throughout the tech and coding industry, and is a powerful tool that can aid coders in daily mundane tasks–boosting productivity–there are significant potential issues that companies need to keep in mind when beginning GenAI integration. While many of these issues have not–and may not–come to fruition, it’s still vital for companies to keep an eye on legal uncertainties when using new tools and technologies, as while these tools may save money today, it could cost them much more than anticipated later on if a product is denied copyright eligibility or if the company is deemed liable for copyright infringement due to said tool use. In an industry that thrives on creative innovation, companies and coders must strike a balance between leveraging AI for efficiency and ensuring they do not undermine the very processes that drive technological progress.

Works Cited

Bousquette, I. (2025, March 4). “How AI Tools Are Reshaping the Coding Workforce”. The Wall Street Journal. https://www.wsj.com/articles/how-ai-tools-are-reshaping-the-coding-workforce-6ad24c86

Hagen, A. (2025, March 2). “GEA1: Generative AI in Educational Spaces” Hagen // Analytics. https://hagenanalytics.com/2025/03/02/gea1-generative-ai-in-educational-spaces/

OpenAI. (2025). ChatGPT (4o) [Large language model]. https://chat.openai.com/

Pandas. (2025). “About Pandas”. Pandas. https://pandas.pydata.org/about/

Thaler v. Perlmutter [2023] No. 22-1564 (BAH) https://ecf.dcd.uscourts.gov/cgi-bin/show_public_doc?2022cv1564-24

Tremayne-Pengelly, A. (2024, July 19). “A.I. Companies Are Running Out of Training Data: Study” Observer. https://observer.com/2024/07/ai-training-data-crisis/

U.S. Copyright Office, Library of Congress. (2023, March 16). Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence. Federal Register. https://www.federalregister.gov/documents/2023/03/16/2023-05321/copyright-registration-guidance-works-containing-material-generated-by-artificial-intelligence