The National Institute of Standards and Technology's Center for AI Standards and Innovation, or CAISI, has signed agreements with Google DeepMind, Microsoft and xAI to support frontier artificial intelligence testing and research tied to national security efforts.
As government and industry leaders expand efforts to evaluate frontier AI systems for national security applications, discussions around AI’s growing role in cybersecurity continue to gain momentum. The 2026 Cyber Summit on May 21 will feature a panel discussion about the role of AI in cyber defense. Reserve your spot now!
NIST said Tuesday the agreements build on previously announced partnerships that were renegotiated to align with directives from Commerce Secretary Howard Lutnick and the White House’s AI Action Plan.
How Could CAISI Agreements Advance AI Assessments?
According to NIST, the agreements allow government evaluators to assess frontier AI models before public release and conduct additional testing after deployment.
CAISI has completed more than 40 evaluations to date, including assessments involving unreleased AI models. Developers frequently provide models with reduced or removed safeguards to support evaluations focused on national security-related capabilities and risks.
The agreements also support testing in classified environments and enable participation from evaluators across government agencies through the TRAINS Taskforce, a group of interagency experts focused on AI-related national security issues.
CAISI Director Chris Fal said independent measurement science plays an important role in understanding frontier AI and related national security implications.
“These expanded industry collaborations help us scale our work in the public interest at a critical moment,” Fall added.
What Is CAISI?
CAISI is a component of the Department of Commerce’s NIST that serves as industry’s primary liaison within the U.S. government to facilitate testing, collaborative research and development of best practices related to commercial AI tools.
In March, CAISI teamed up with the nonprofit OpenMined to develop methods for evaluating AI systems while preserving data confidentiality. It also partnered with the General Services Administration to establish approaches to AI systems evaluation.
The organization also introduced the AI Agent Standards Initiative to ensure interoperable and secure adoption of agentic AI systems and sought public input on draft best practices for automated AI benchmark testing.
