NIST Partners with Anthropic and OpenAI for AI Safety Research

Gaithersburg, Maryland—Today, the U. S. Department of Commerce's National Institute of Standards and Technology (NIST) announced collaborative partnerships with Anthropic and OpenAI, formalizing their joint efforts in AI safety research, testing, and evaluation. Under the Memorandum of Understanding, the U. S. AI Safety Institute will receive early access to major new models from both companies, allowing for collaborative research on evaluating capabilities and addressing safety risks associated with AI. The agreements also facilitate the sharing of insights on potential safety improvements to Anthropic and OpenAI's models, in coordination with the U. K. AI Safety Institute. According to Elizabeth Kelly, Director of the U. S.

AI Safety Institute, these agreements mark an important milestone in advancing the science of AI safety and promoting responsible innovation. The aim is to support the development and stewardship of AI with a focus on safety. The evaluations conducted through these partnerships contribute to NIST's rich history of advancing technology, standards, and measurement science. By fostering deep collaboration and exploratory research, NIST aims to address the various risks associated with advanced AI systems. Furthermore, these evaluations align with the Biden-Harris administration's Executive Order on AI and the commitments made by leading AI model developers to ensure the safe, secure, and trustworthy development and use of AI. About the U. S. AI Safety Institute: