US Center for AI Standards and Innovation (CAISI, formerly US AI Safety Institute)
NIST body serving as industry's point of contact for voluntary AI testing, standards and national-security evaluations.
The US government's AI testing centre, created under Biden and rebranded under Trump. It now tests all major US labs' models voluntarily and benchmarks Chinese models.
Focus
- Safety evaluations
- Transparency
Funding, as far as we know
US government (NIST / Department of Commerce) - https://www.nist.gov/caisi
What they've actually done
Signed pre-deployment testing agreements with OpenAI and Anthropic (as US AISI)
29 August 2024
Renamed CAISI under the Trump administration, dropping 'safety' and emphasising standards and adversary-model evaluation
2025-06
Expanded voluntary pre-release security reviews to Google, Microsoft and xAI
5 May 2026
Published evaluations of PRC models DeepSeek V4 Pro (May 2026) and Z.ai GLM-5.2 (July 2026)
2026