USPTO: Harnessing AI for Enhanced Patent Protection While Valuing Human Expertise

The U.S. Patent and Trademark Office (USPTO) has outlined new regulations indicating that a human must make a "significant contribution" when utilizing AI tools in the development of patents. The long-anticipated guidance clarifies that while applicants can harness AI technologies like ChatGPT during the invention process, these tools should serve primarily as assistants and not be responsible for the invention's entire creation.

In this guidance, the USPTO acknowledges the role of AI but stipulates that human involvement is essential to secure patent rights. This regulatory framework will officially take effect on February 13. Kathi Vidal, the USPTO director, emphasized in her blog post that patent protection can be pursued for inventions with substantial human input. To qualify for a patent, at least one human must be named as an inventor and demonstrate significant contributions to the invention.

Vidal indicated that the USPTO would generally assume that named inventors are indeed the true creators, only raising questions based on official records or supporting evidence. She elaborated that the role of AI or other advanced systems in the invention process will not hinder patent eligibility; rather, the key factor is whether the contributions of a human inventor can be clearly identified in the application.

Moreover, the guidance specifies that even a significant contribution through something as simple as crafting a prompt could suffice. However, merely having "intellectual domination" over an AI system will not automatically qualify someone as an inventor of an invention generated by that system.

Jamie Nafziger, a partner at Dorsey & Whitney, praised the USPTO's sensible approach, stating, “The challenge will be in implementing the guidance. How sophisticated must a prompt be for a specific invention? Additionally, what level of planning is necessary for training an AI system? Patent examiners are sure to face intriguing challenges ahead.”

In line with President Biden’s executive order from last October, the USPTO has been tasked with exploring frameworks related to AI and intellectual property (IP). Alongside these patent regulations, the office has been directed to collaborate with the U.S. Copyright Office, aiming to provide recommendations on potential executive actions that address copyright issues arising from AI, including the extent of protection for AI-generated works and the implications of using copyrighted materials for AI training.

Intellectual property rights have emerged as a critical concern in the AI landscape. Companies are wary of disclosing sensitive information to AI systems, reminiscent of incidents like Samsung's data breach last year. Meanwhile, developers face challenges related to accessing copyrighted materials, which could stifle the innovation of new models. For instance, OpenAI has noted that training models without utilizing copyrighted content would be “impossible.”

It's important to note that AI cannot hold IP rights. Courts in the U.K., Europe, and the U.S. have consistently ruled against granting patent protections to inventions where AI is designated as the inventor. This trend extends to copyright claims, as seen in U.S. courts rejecting protections for AI-generated artworks, including a piece by Midjourney that won a Colorado State Fair art contest.

Randy McCarthy from Hall Estill advises applicants to meticulously document their processes and ensure substantial human input in the creation of their inventions. By doing so, they can help validate their intellectual property claims and navigate the evolving landscape of AI and innovation with greater confidence.