Our sensitivity to adoption includes intellectual property ownership and ability to copyright our work.
Quick recap: As companies increasingly deploy AI-powered tools, developers are turning to large language models to streamline tasks like code generation and test case development. However, a crucial concern remains: who holds the intellectual property rights to AI-generated content?
Legal precedent: Last year, the U.S. District Court ruled that copyright law mandates human authorship, making AI-generated works ineligible for protection. The decision, highlighted in Thaler v. Perlmutter, came after an AI developer sought to copyright a work created solely by an algorithm without human involvement. The court emphasized that under U.S. copyright law, only works created by human authors can receive protection, leaving AI-generated content in a legal gray area.
Experts weigh in: The ruling has significant implications for businesses and creators increasingly relying on AI to produce content. “Our sensitivity to adoption includes intellectual property ownership and ability to copyright our work,” says Sean John Connelly, lead machine learning software engineer at WideSense, a company that provides solutions for the design, deployment, and profitable operation of electric vehicle fleets. “The USPTO doesn’t permit copyright of ‘non-human’ generated material. Therefore, it is unlikely that content generated solely by an AI can become a company's intellectual property.”
The lack of clarity surrounding AI-generated work has also left many companies questioning where the boundaries lie when humans use LLMs to assist in generating code, says Connelly. “There is little legal precedence when a human utilizes an LLM to augment their code—where is the line drawn? This will undoubtedly be coming to a court or legislative branch near you soon.”
Evaluating LLMs for code ownership assurance: Connelly’s team is currently testing various LLMs to support product development, and intellectual property assurance plays a significant role in their decision-making. “In regards to code generation output, Claude and OpenAI appear to provide the most assurance that the user ‘owns’ or has rights to the output in their terms and conditions, while Gemini and Jasper provide the least assurance from what they state or do not explicitly state,” Connelly said.
The bottom line: IP ownership is emerging as one of the most pressing challenges across industries. As legal frameworks struggle to keep pace with rapid advancements, companies and developers must scrutinize the terms governing AI-generated content, ensuring alignment between LLM usage and long-term intellectual property strategies.
