USA 
United Kingdom
The US Patent and Trademark Office banned employees from using generative artificial intelligence (generative-AI) in 2023, a memo obtained by WIRED has revealed.
The memo cited concerns over “bias, unpredictability, and malicious behavior,” prohibiting workers from using any AI-generated text, images, and videos.
There has been progress in the 18 months since the order was issued. Paul Fucito, a press secretary for the USPTO, clarified to WIRED that employees are now able to use some generative-AI tools—but only “state-of-the-art” models available the agency’s internal testing environment.
USPTO working to bring “responsible” gen-AI
The USPTO is responsible for awarding patents, registering trademarks, and protecting inventors. It sits within the Department of Commerce.
In the original memo sent to agency staff, Jamie Hulcombe, Chief Information Officer at the USPTO told employees the office was “committed to pursuing innovation,” though was still “working to bring these capabilities to the office in a responsible way.”
“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,” the memo continued.
In the same month, Scott Beliveau, Chief of Advanced Analytics at the USPTO, told Forbes he had banned employees from using LLMs. “We started mitigating the risks of LLMs [large language models] by prohibiting our employees and contractors from using generative AI tools,” he explained. “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.”
Testing environments helping Patent Office innovate with AI
The “responsible” approach has helped the USPTO make progress within its AI-testing environment.
Fucito noted that as of November 2024, “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.”
Other organizations embracing AI adoption have relied heavily on test environments. Mary Alice Vuicic, CPO at Thomson Reuters told HR Grapevine in an interview this week about the launch of “Open Arena,” an internal playground that gives the company’s 26,000 employees access to test four LLM-enabled generative AI tools.
USPTO staff are still prohibited from using generative AI tools outside of the test environment, including OpenAI tool ChatGPT, for completing day-to-day work.
While the use of generative-AI tools come with strict guidance, agency staffers are free to use approved AI-powered tools in their work.
Navigating Global Workforce Management: A Guide for Businesses Expanding Globally
Expanding your business internationally is more accessible than ever, but managing a global workforce presents unique challenges. Our guide helps you choose the right provider, avoid common mistakes, and identify essential features for scaling your business across the globe.
This guide covers:
Choosing the Right Provider: Key factors like scalability, compliance, and cost-effectiveness.
Avoiding Common Pitfalls: How to prevent frequent selection errors.
Critical Features: Must-haves like international payroll, compliance, and benefits support.
Compliance & Risk Tips: Ensuring your provider handles local laws and regulations.
Scalable Solutions: How the right provider supports long-term growth.
Download your free guide to start your global expansion today!
For example, the USPTO has introduced an AI-based tool dubbed the ‘Similarity Search’ feature. The tool allows workers – known as ‘examiners’ – to input patent application information and uses AI models to search its database for similar existing applications.
Agency documentation about Similarity Search reports it provides “significantly faster results for examiners.”
“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,” the agency said.
The USPTO noted the feature has been designed to mitigate “potential model biases” and to meet National Institute of Standards and Technology (NIST) security controls.
USPTO Chief: Government bureaucracy slows down innovation
In 2023, Holcombe criticized the amount of red tape in government agencies at an event sponsored by Google.
“Everything we do in the government is pretty stupid, when you compare it to the commercial world, right?” the Chief Information Officer stated.
He suggested that compliance processes, as well as drawn-out technology procurement cycles and strict budgets all slowed down innovation and AI adoption.
Other government agencies have been encouraging and banning generative AI use across a variety of different applications.
Per FedScoop reports, The National Aeronautics and Space Administration (NASA) has prohibited staff from uploading sensitive data to AI chatbots; but is using AI to write code, summarize research, and to power a chatbot that it hopes will “democratize access to spaceborne data.”
Meanwhile, according to a 404 Media, the National Archives and Records Administration banned employees from using ChatGPT on their work laptops; but encouraged them to treat Google Gemini as “a co-worker,” despite worker fears about the technology’s accuracy.
Previous reports published by the USPTO and other departments like the Department of Defense have suggested AI adoption is being slowed by insufficient skills and budgets.
