The U.S. Patent and Trademark Office (USPTO) banned staff from using generative AI tools like ChatGPT and Anthropic’s Claude in 2023, citing security concerns and risks associated with bias, unpredictability, and malicious behavior. USPTO’s Chief Information Officer Jamie Holcombe emphasized the agency’s commitment to innovation but stressed the need for responsible integration of AI technologies. Despite the ban, the USPTO continues to explore generative AI applications within controlled settings, ensuring they align with the agency’s goals and responsibilities. This decision was detailed in an internal memo obtained by WIRED via a public records request.
The Scope of the Ban
Under the directive, USPTO employees are prohibited from using generative AI tools for work tasks or relying on their outputs, including text, images, and videos. The memo also outlined the potential dangers of these technologies, from inaccuracies in critical patent data to broader ethical implications.
However, generative AI is not entirely off-limits at the USPTO. Employees can experiment with these tools in the agency’s AI Lab, a controlled testing environment designed to assess AI’s potential and limitations.
“Innovators from across the USPTO are now using the AI Lab to better understand generative AI’s capabilities and prototype AI-powered solutions to critical business needs,” said USPTO’s press secretary Paul Fucito.
AI in Patent Examination
While banning generative AI for routine tasks, the USPTO leverages artificial intelligence in other ways. Its internal tools, such as the Patents End-to-End Search system, utilize machine learning to help examiners perform prior art searches. These features, like “More Like This Document” and “Similarity Search,” aid in the review of millions of patent applications annually.
The agency also acknowledges the growing role of AI in patent submissions. Although applicants are not required to disclose their use of generative AI, tools like ChatGPT have been employed to draft patent documents. The USPTO is working to adapt its systems to handle these emerging trends responsibly.
Generative AI Across Government Agencies
The USPTO is not alone in navigating generative AI. Other agencies, such as NASA and the Department of Energy, are exploring AI’s potential while grappling with funding and resource challenges. The National Archives and Records Administration has taken a mixed approach, banning ChatGPT on work devices while developing AI-powered chatbots for public access.
The inconsistent adoption across federal agencies reflects broader uncertainties about AI’s risks and benefits. For the USPTO, however, its ban is a clear step to protect its critical mission while laying the groundwork for future innovations.
As Holcombe stated, “The USPTO recognizes the need, through technical mitigations and human governance, to cabin the risks arising from the use of AI in practice.”