Pacific Northwest National Laboratory (PNNL) has joined the initial Testing Risks of AI for National Security (TRAINS) Taskforce to address national security concerns related to artificial intelligence (AI). PNNL is one of ten national laboratories to join the taskforce alongside representation from across the U.S. government, including the Department of Commerce, Department of Defense, Department of Homeland Security, and the National Institutes of Health. 

“The TRAINS taskforce provides an opportunity for AI safety experts to conduct evaluations of AI models before they are deployed,” said Jessica Baweja, applied research psychologist at PNNL. “We can help determine what risks, if any, the models might present to the public and if they could be used for harmful activities.”

Baweja leads the AI literacy subcommittee within the Center for AI @PNNL. The Center plays a pivotal role in the use and development of safe, secure, and trustworthy AI at PNNL. In 2024, the Center supported the development of an AI literacy accelerator for the Battelle Digital Resource Center, an information sharing platform used by the Battelle-managed national laboratories. The Center’s subcommittees on literacy and data governance promote generative AI best practices and robust AI-ready data generation and management frameworks.  

“PNNL researchers have worked hard to integrate AI into every part of our mission,” said Nathan Hodas, who leads the Generative AI Investment at PNNL. Hodas will join Baweja in representing PNNL on the TRAINS taskforce.

“With rapidly evolving technologies, such as AI, there’s often not enough time to establish best practices or fully understand their boundaries and capabilities before adopting them,” said Hodas. “With this taskforce, we can explore the strengths and weakness of new AI to inform which safeguards may be needed for these technologies.”

The U.S. AI Safety Institute—part of the National Institute of Standards and Technology within the Department of Commerce—will chair the TRAINS taskforce. This effort represents a whole-government approach toward AI safety, as directed by the National Security Memorandum on AI.