The General Services Administration (GSA) is partnering with the National Institute of Standards and Technology (NIST) to improve how federal agencies evaluate artificial intelligence (AI) tools prior to deployment. This collaboration aims to streamline AI adoption across government while addressing concerns about the technology’s security and reliability.
Announced by GSA, the partnership will leverage the expertise of NIST’s Center for AI Standards and Innovation (CAISI), previously known as the AI Safety Institute. The effort is designed to support GSA’s USAi platform, which allows federal agencies to test and experiment with various AI models. According to GSA, this relationship will enable faster adoption of AI tools while ensuring confidence in their use.
"By combining GSA’s government-wide reach with NIST’s AI evaluation expertise, we’re strengthening how the federal government deploys AI", said GSA Administrator Edward Forst in the agency’s announcement.
Expanding AI Capabilities Across Federal Agencies
Federal agencies have been increasingly adopting AI tools to assist workers, automate administrative tasks, and address operational challenges. However, as AI technology evolves, concerns about its reliability and security remain at the forefront of deployment strategies. To address these issues, CAISI will provide tools and guidance to help GSA evaluate advanced AI models, interpret benchmarks, and conduct real-world testing in federal workflows.
The partnership also plans to produce "guidelines and checklists" for other federal agencies, ensuring a standardized approach to evaluating and deploying AI systems.
"We’re at a pivotal time in the AI revolution and this partnership between CAISI and GSA will enable federal agencies to adopt AI in ways that help the American people", said Craig Burkhardt, NIST’s acting director, in a statement included with the announcement.
Aligning with National AI Policies
This collaboration aligns with recommendations from President Donald Trump’s "AI Action Plan", released last year, which called for establishing an "evaluations ecosystem" to provide resources for agencies conducting AI testing. The new partnership supports this initiative by creating tools and frameworks that will uphold rigorous standards for AI in government use.
Additionally, the GSA is working on draft contract language for AI system deployments under its Multiple Awards Schedule. The draft clause outlines rules for intellectual property rights, government data usage, security incident reporting, and government evaluation rights. It also specifies that assessments will examine factors such as bias, truthfulness, safety, and ideological content.
Addressing Broader Challenges in AI Oversight
The announcement did not address whether the evaluations will explicitly focus on "ideological biases" or "social agendas", as targeted in Trump’s executive order aimed at preventing so-called "woke AI" in government. Furthermore, a GSA spokesperson did not respond to requests for details on the current evaluation process, assessment rubrics, or plans to make their findings public.
While questions remain, this collaboration marks a significant step in advancing the government’s ability to adopt AI responsibly and effectively. By leveraging the combined strengths of GSA and NIST, federal agencies have the potential to deploy AI solutions that are both innovative and secure.
