Last year, the US Patent and Trademark Office banned the use of AI tools like ChatGPT in the office. An internal memo obtained by WIRED through a public records request detailed the ins and outs of the ban. But that doesn’t mean the Patent Office isn’t using AI at all—quite the opposite.
According to WIRED, USPTO chief technology officer Jamie Holcombe sent out a memo to Patent Office employees in April 2023. It prohibited workers from using generative AI tools for work due to the system’s “bias, unpredictability, and malicious behavior.” Holcombe said that the Patent Office is “committed to pursuing innovation with our agency” and “working to bring these capabilities to the office in a responsible way.”
That same month, Scott Beliveau—the USPTO’s Branch Chief of Advanced Analytics—parroted the language of Holcombe’s memo in an interview with Forbes. “We are as committed to pursuing innovation within our agency,” Beliveau told Forbes when asked about how the Patent Office was working to minimize the risks of AI.
Beliveau told Forbes that his office wasn’t allowed to use LLMs at all. “We started mitigating the risks of LLMs by prohibiting our employees and contractors from using generative AI tools,” he said. “This immediate action was taken while we continue to explore ways to bring LLM capabilities to the agency in a responsible manner that serves America’s innovators.”
This does not mean that the Patent Office is ignoring AI. The Patent Office examines millions of legal documents every year and it has admitted it’s using AI tools to help with the work. “Patent examiners are performing AI-enabled prior art searches using features like More Like This Document (MLTD) and Similarity Search, in the Office’s Patents End-to-End (PE2E) Search tool,” it said in a document detailing its exhaustive use of AI published in the Federal Register. “Patent practitioners are increasingly relying on AI-based tools to research prior art, automate the patent application review process, and to gain insights into examiner behavior.”
In addition to its own use of AI, the Patent Office knows that the genie is out of the bottle regarding generative AI submissions to its office. “The capabilities of these tools continue to grow, and there is no prohibition against using these computer tools in drafting documents for submission to the USPTO,” it said in the Federal Register document. “Nor is there a general obligation to disclose to the USPTO the use of such tools.”
The Patent Office’s press secretary also told WIRED that USPTO employees can use generative AI tools within an internal testing environment. “Innovators from across the USPTO are now using the AI Lab to better understand generative AI’s capabilities and limitations and to prototype AI-powered solutions to critical business needs,” the USPTO told WIRED.
Government policies regarding the use of generative AI are all over the place. The USPTO and the Department of Defense are both forward-thinking, compared to many of their peers. In September, multiple agencies published reports on how they’re using AI, and the results weren’t great. Adoption is slow across the board. The agencies don’t have enough money or talent to develop their own tools. The Department of Energy, the agency in charge of America’s nuclear weapons, whined that it didn’t have enough graphics cards.
Meanwhile, the Patent Office has been plugging away at this problem for years. It rolled out its own internal search tool several years ago. “While the USPTO is committed to maximizing AI’s benefits and seeing them distributed broadly across society, the USPTO recognizes the need, through technical mitigations and human governance, to cabin the risks arising from the use of AI in practice before the USPTO,” it said in its guidance memo.