The China-based artificial intelligence model DeepSeek isn’t available for widespread use at the Department of Energy, but approval of some elements may be possible following a study by two of its national labs, an agency IT official said Tuesday.
DeepSeek’s launch has prompted congressional proposals to rein in its use in government and proactive bans by several federal agencies, including DOE. But during a panel at a FedScoop-produced Salesforce event, Bridget Carper — the agency’s deputy CIO for architecture, engineering, technology and innovation — said the model has still been studied by two DOE national labs.
Carper said the agency allowed two of its labs — which she didn’t identify — to look at the system “because there’s value in testing the open models. There’s value in understanding the performance. How does it actually compare?”
The separate labs looked at the model to see if they could do comparisons with alternatives they had, Carper said. Those studies also took place with guardrails. They were controlled, sanctioned and fully documented, she said. And ultimately, they found some potential benefits.
“We did learn about a couple of the things that we thought were positive, that we could leverage of the model that we may approve,” Carper said.
Her comments came in response to a question about the energy agency’s approach to exploring new AI technologies. Carper explained that DOE is open to anything and everything, albeit with some restrictions.
“We have no off limits across the board. However, there are limitations in who is capable of doing it,” she said. DeepSeek, for example, is not able to be downloaded at the agency and staff are prohibited from inputting DOE data.
DeepSeek’s release of its R1 model earlier this year put U.S.-based platforms on notice, boasting lower costs, better efficiency, and transparent reasoning. On Tuesday, the company released its V3.1 model, which aims to further enhance its capabilities.
As FedScoop previously reported, there isn’t much appetite to access DeepSeek on government systems, but there is at least one instance of an attempt. The U.S. Department of Agriculture blocked attempted access to the site, which is banned along with other public AI platforms.
Written by Madison Alder
Madison Alder is a reporter for FedScoop in Washington, D.C., covering government technology. Her reporting has included tracking government uses of artificial intelligence and monitoring changes in federal contracting. She’s broadly interested in issues involving health, law, and data. Before joining FedScoop, Madison was a reporter at Bloomberg Law where she covered several beats, including the federal judiciary, health policy, and employee benefits. A west-coaster at heart, Madison is originally from Seattle and is a graduate of the Walter Cronkite School of Journalism and Mass Communication at Arizona State University.